Saved in:
Bibliographic Details
Main Authors: Berrayana, Lina, Heakl, Ahmed, Sohail, Abdullah, Hofmann, Thomas, Khan, Salman, Chen, Wei
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.09184
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908875837407232
author Berrayana, Lina
Heakl, Ahmed
Sohail, Abdullah
Hofmann, Thomas
Khan, Salman
Chen, Wei
author_facet Berrayana, Lina
Heakl, Ahmed
Sohail, Abdullah
Hofmann, Thomas
Khan, Salman
Chen, Wei
contents Most multi-agent systems rely exclusively on autoregressive language models (ARMs) that are based on sequential generation. Although effective for fluent text, ARMs limit global reasoning and plan revision. On the other hand, Discrete Diffusion Language Models (DDLMs) enable non-sequential, globally revisable generation and have shown strong planning capabilities, but their limited text fluency hinders direct collaboration with ARMs. We introduce Latent-DARM, a latent-space communication framework bridging DDLM (planners) and ARM (executors), maximizing collaborative benefits. Across mathematical, scientific, and commonsense reasoning benchmarks, Latent-DARM outperforms text-based interfaces on average, improving accuracy from 27.0% to 36.0% on DART-5 and from 0.0% to 14.0% on AIME2024. Latent-DARM approaches the results of state-of-the-art reasoning models while using less than 2.2% of its token budget. This work advances multi-agent collaboration among agents with heterogeneous models.
format Preprint
id arxiv_https___arxiv_org_abs_2603_09184
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Latent-DARM: Bridging Discrete Diffusion And Autoregressive Models For Reasoning
Berrayana, Lina
Heakl, Ahmed
Sohail, Abdullah
Hofmann, Thomas
Khan, Salman
Chen, Wei
Machine Learning
Artificial Intelligence
Most multi-agent systems rely exclusively on autoregressive language models (ARMs) that are based on sequential generation. Although effective for fluent text, ARMs limit global reasoning and plan revision. On the other hand, Discrete Diffusion Language Models (DDLMs) enable non-sequential, globally revisable generation and have shown strong planning capabilities, but their limited text fluency hinders direct collaboration with ARMs. We introduce Latent-DARM, a latent-space communication framework bridging DDLM (planners) and ARM (executors), maximizing collaborative benefits. Across mathematical, scientific, and commonsense reasoning benchmarks, Latent-DARM outperforms text-based interfaces on average, improving accuracy from 27.0% to 36.0% on DART-5 and from 0.0% to 14.0% on AIME2024. Latent-DARM approaches the results of state-of-the-art reasoning models while using less than 2.2% of its token budget. This work advances multi-agent collaboration among agents with heterogeneous models.
title Latent-DARM: Bridging Discrete Diffusion And Autoregressive Models For Reasoning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2603.09184