SOURCE-LINKED INTELLIGENCE

F$^{2}$DR: A Fine-Grained Full-Pipeline Reward Framework for DeepSearch Workflows

arXiv · AI, language, vision and robotics · article · Sep 17, 2026 · UTC

With the widespread industrial deployment of Large Language Models (LLMs), DeepSearch has emerged as the dominant paradigm for resolving complex user queries. It typically operates through an iterative closed-loop workflow consisting of planning and reflection, information retrieval, and answer generation. However, existing reward models (RMs) and evaluation benchmarks are primarily designed for static single-turn tasks, failing to capture the full-pipeline complexity of DeepSearch workflows. To address this limitation, we propose F2DR, a fine-grained full-pipeline DeepSearch reward framework.

Read original source ↗ Open in workspace

recordType

paper

region

Global