Exploring Reasoning Reward Model for Agents

TL;DR

This paper introduces Agent-RRM, a multi-faceted reasoning reward model that provides structured feedback to improve agentic reinforcement learning, leading to significant performance improvements across multiple benchmarks.

Contribution

The paper proposes a novel reasoning reward model with explicit reasoning traces and critique, and demonstrates its effectiveness through three integration strategies in diverse benchmarks.

Findings

Reagent-U achieves 43.7% on GAIA and 46.2% on WebWalkerQA.

Structured feedback improves agent reasoning and performance.

Extensive evaluations validate the effectiveness of the proposed reward model.

Abstract

Agentic Reinforcement Learning (Agentic RL) has achieved notable success in enabling agents to perform complex reasoning and tool use. However, most methods still relies on sparse outcome-based reward for training. Such feedback fails to differentiate intermediate reasoning quality, leading to suboptimal training results. In this paper, we introduce Agent Reasoning Reward Model (Agent-RRM), a multi-faceted reward model that produces structured feedback for agentic trajectories, including (1) an explicit reasoning trace , (2) a focused critique that provides refinement guidance by highlighting reasoning flaws, and (3) an overall score that evaluates process performance. Leveraging these signals, we systematically investigate three integration strategies: Reagent-C (text-augmented refinement), Reagent-R (reward-augmented guidance), and Reagent-U (unified feedback integration). Extensive evaluations across 12 diverse benchmarks demonstrate that Reagent-U yields substantial performance leaps, achieving 43.7% on GAIA and 46.2% on WebWalkerQA, validating the effectiveness of our reasoning reward model and training schemes. Code, models, and datasets are all released to facilitate future research.