C$^2$DLM: Causal Concept-Guided Diffusion Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Han, Kairong, Shan, Nuanqiao, Zhao, Ziyu, Hu, Zijing, Dong, Xinpeng, Ye, Junjian, Pan, Lujia, Wu, Fei, Kuang, Kun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912732585918464
author Han, Kairong
Shan, Nuanqiao
Zhao, Ziyu
Hu, Zijing
Dong, Xinpeng
Ye, Junjian
Pan, Lujia
Wu, Fei
Kuang, Kun
author_facet Han, Kairong
Shan, Nuanqiao
Zhao, Ziyu
Hu, Zijing
Dong, Xinpeng
Ye, Junjian
Pan, Lujia
Wu, Fei
Kuang, Kun
contents Autoregressive (AR) language models and Diffusion Language Models (DLMs) constitute the two principal paradigms of large language models. However, both paradigms suffer from insufficient reasoning capabilities. Human reasoning inherently relies on causal knowledge and thought, which are reflected in natural language. But in the AR paradigm, language is modeled as next token prediction (a strictly left-to-right, token-by-token order), whereas natural language itself exhibits more flexible causal structures. In the DLM paradigm, the attention mechanism is fully connected, which entirely disregards causal order. To fill this gap, we propose a \underline{\textbf{C}}ausal \underline{\textbf{C}}oncept-Guided \underline{\textbf{D}}iffusion \underline{\textbf{L}}anguage \underline{\textbf{M}}odel (C$^2$DLM). Starting from DLM's fully connected attention, C$^2$DLM first obtains a concept-level causal graph from the teacher model, and then explicitly guides attention to learn causal relationships between concepts. By focusing on causal relationships and avoiding interference from difficult subgoals involving causal inversion, C$^2$DLM improves 12\% with about 3.2 times training speedup in the COT-OrderPerturb task, and achieves an average gain of 1.31\% across six downstream reasoning tasks. More details in the repository ~\href{https://github.com/Kairong-Han/C-2-DLM}{here}.
format Preprint
id arxiv_https___arxiv_org_abs_2511_22146
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle C$^2$DLM: Causal Concept-Guided Diffusion Large Language Models
Han, Kairong
Shan, Nuanqiao
Zhao, Ziyu
Hu, Zijing
Dong, Xinpeng
Ye, Junjian
Pan, Lujia
Wu, Fei
Kuang, Kun
Computation and Language
Autoregressive (AR) language models and Diffusion Language Models (DLMs) constitute the two principal paradigms of large language models. However, both paradigms suffer from insufficient reasoning capabilities. Human reasoning inherently relies on causal knowledge and thought, which are reflected in natural language. But in the AR paradigm, language is modeled as next token prediction (a strictly left-to-right, token-by-token order), whereas natural language itself exhibits more flexible causal structures. In the DLM paradigm, the attention mechanism is fully connected, which entirely disregards causal order. To fill this gap, we propose a \underline{\textbf{C}}ausal \underline{\textbf{C}}oncept-Guided \underline{\textbf{D}}iffusion \underline{\textbf{L}}anguage \underline{\textbf{M}}odel (C$^2$DLM). Starting from DLM's fully connected attention, C$^2$DLM first obtains a concept-level causal graph from the teacher model, and then explicitly guides attention to learn causal relationships between concepts. By focusing on causal relationships and avoiding interference from difficult subgoals involving causal inversion, C$^2$DLM improves 12\% with about 3.2 times training speedup in the COT-OrderPerturb task, and achieves an average gain of 1.31\% across six downstream reasoning tasks. More details in the repository ~\href{https://github.com/Kairong-Han/C-2-DLM}{here}.
title C$^2$DLM: Causal Concept-Guided Diffusion Large Language Models
topic Computation and Language
url https://arxiv.org/abs/2511.22146