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

2026-02-03

AI Summary

This work addresses the high latency inherent in the ReAct paradigm due to its serial execution and the limited tool-calling and reasoning capabilities of diffusion-based large language models (dLLMs). To overcome these challenges, the authors propose DLLM-Searcher, a novel framework enhanced through a two-stage post-training process: Agentic Supervised Fine-Tuning (SFT) and Variance-Reduced Preference Optimization (VRPO), which collectively improve the model’s retrieval and reasoning proficiency. Furthermore, they introduce Parallel ReAct (P-ReAct), a pioneering paradigm that enables the model to continue generating subsequent reasoning steps while awaiting tool responses. Experimental results demonstrate that DLLM-Searcher achieves performance on par with state-of-the-art LLM-based search agents, while P-ReAct delivers approximately 15% end-to-end inference acceleration.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

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

Problem

• Diffusion Large Language Model
• Search Agents
• Latency Challenge
• Agent Ability Challenge
• ReAct paradigm

Innovation

• Diffusion Large Language Model
• Search Agent
• Parallel Reasoning
• Tool Calling
• Preference Optimization