Papers
arxiv:2602.07035

DLLM-Searcher: Adapting Diffusion Large Language Model for Search Agents

Published on Feb 3
ยท Submitted by
SunZX
on Feb 11
Authors:
,
,
,
,
,

Abstract

Diffusion Large Language Models are optimized for search agents through enhanced reasoning capabilities and reduced latency via a parallel reasoning paradigm.

AI-generated summary

Recently, Diffusion Large Language Models (dLLMs) have demonstrated unique efficiency advantages, enabled by their inherently parallel decoding mechanism and flexible generation paradigm. Meanwhile, despite the rapid advancement of Search Agents, their practical deployment is constrained by a fundamental limitation, termed as 1) Latency Challenge: the serial execution of multi-round reasoning, tool calling, and tool response waiting under the ReAct agent paradigm induces severe end-to-end latency. Intuitively, dLLMs can leverage their distinctive strengths to optimize the operational efficiency of agents under the ReAct agent paradigm. Practically, existing dLLM backbones face the 2) Agent Ability Challenge. That is, existing dLLMs exhibit remarkably weak reasoning and tool-calling capabilities, preventing these advantages from being effectively realized in practice. In this paper, we propose DLLM-Searcher, an optimization framework for dLLM-based Search Agents. To solve the Agent Ability Challenge, we design a two-stage post-training pipeline encompassing Agentic Supervised Fine-Tuning (Agentic SFT) and Agentic Variance-Reduced Preference Optimization Agentic VRPO, which enhances the backbone dLLM's information seeking and reasoning capabilities. To mitigate the Latency Challenge, we leverage the flexible generation mechanism of dLLMs and propose a novel agent paradigm termed Parallel-Reasoning and Acting P-ReAct. P-ReAct guides the model to prioritize decoding tool_call instructions, thereby allowing the model to keep thinking while waiting for the tool's return. Experimental results demonstrate that DLLM-Searcher achieves performance comparable to mainstream LLM-based search agents and P-ReAct delivers approximately 15% inference acceleration. Our code is available at https://anonymous.4open.science/r/DLLM-Searcher-553C

Community

Paper author Paper submitter

๐Ÿง ๐Ÿ” DLLM-Searcher: Adapting Diffusion Large Language Models for Search Agents

Diffusion Large Language Models (dLLMs) offer flexible generation but struggle as search agents due to latency and weak tool-use capabilities. This paper introduces DLLM-Searcher, a framework that adapts dLLMs for efficient, agentic search and retrieval.

๐Ÿš€ Key ideas:

  • Parallel-Reasoning and Acting (P-ReAct):
    Enables parallel reasoning and tool execution using diffusionโ€™s non-autoregressive generation, significantly reducing inference latency.
  • Agent-oriented post-training:
    A two-stage pipeline with Agentic Supervised Fine-Tuning (SFT) + Agentic Variance-Reduced Preference Optimization (VRPO) improves reasoning structure, tool calling, and search reliability.

๐Ÿ“Š Results:

  • Competitive performance with strong autoregressive LLM-based search agents on multi-hop retrieval tasks
  • Up to ~15% speedup in end-to-end inference with P-ReAct

๐Ÿ’ก Why it matters:
DLLM-Searcher shows that diffusion LLMs can be practical and efficient search agents, opening a new direction for low-latency, agentic information retrieval systems.

Sign up or log in to comment

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2602.07035 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2602.07035 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2602.07035 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.