RSPO: Reward-Swap Policy Optimization for Multi-Turn LLM Agents

• URL: http://arxiv.org/abs/2607.04713v1
• Date: Mon, 06 Jul 2026 06:32:39 GMT
• Title: RSPO: Reward-Swap Policy Optimization for Multi-Turn LLM Agents
• Abstract summary: We propose Reward-Swap Policy Optimization (RSPO) to leverage the rich information from dense process rewards to facilitate training with outcome rewards.<n>RSPO ensures the diversity of sampled trajectories while guaranteeing consistency between the optimization objective and the true outcome rewards.
• Score: 24.215311212134882
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Reinforcement learning holds significant potential for training large language models (LLMs) to handle multi-turn interactive tasks. However, in long-horizon, multi-turn tasks characterized by sparse outcome rewards, directly training with outcome rewards often results in slow convergence due to the sparsity of signals and the lack of fine-grained feedback. Furthermore, the model may fail to learn successful trajectories that are not sampled during training, thereby limiting its performance. Conversely, while employing customized dense process rewards provides richer signals and accelerates convergence, these surrogate rewards may exhibit potential misalignment with the ground-truth outcome rewards. This inconsistency can bias the training direction and ultimately degrade the model's final performance. In this work, we propose Reward-Swap Policy Optimization (RSPO), a method designed to leverage the rich information from dense process rewards to facilitate training with outcome rewards. By utilizing a reward-swap mechanism, RSPO ensures the diversity of sampled trajectories while guaranteeing consistency between the optimization objective and the true outcome rewards, thereby elevating the performance ceiling of the model. We conduct extensive experiments on two challenging agent benchmarks, WebShop and ALFWorld. By applying our method to various reinforcement learning algorithms, including GRPO, PPO, and GiGPO, we demonstrate that RSPO achieves consistent performance improvements across different baselines and benchmarks.

Related papers

• Pareto-Guided Optimal Transport for Multi-Reward Alignment [59.776579791864805]
  We show how to optimize against a unified global target under heterogeneous reward upper bounds.<n>We develop both online and offline optimization strategies tailored to diverse reward signal characteristics.<n>Our approach outperforms strong baselines with an 11% gain in Joint Domination Rate (JDR) and achieves a near 80% win rate in human evaluations.
  arXiv  Detail & Related papers  (2026-05-13T08:19:48Z)
• TMPO: Trajectory Matching Policy Optimization for Diverse and Efficient Diffusion Alignment [52.570581883709345]
  We propose Trajectory Matching Policy Balance Optimization (TMPO), which replaces reward with human-level reward distribution matching.<n>TMPO improves generative diversity over state-of-the-art methods by 9.1%, and achieves competitive performance in all downstream and efficiency metrics.<n>To further reduce multi-trajectory training time on large-scale flow- prefixes, TMPO incorporates Dynamic Tree Sampling models, where trajectories share denoising and branch at dynamically scheduled steps.
  arXiv  Detail & Related papers  (2026-05-09T04:41:02Z)
• Information Gain-based Policy Optimization: A Simple and Effective Approach for Multi-Turn LLM Agents [28.145430029174577]
  Large language model (LLM)-based agents are increasingly trained with reinforcement learning (RL) to enhance their ability to interact with external environments.<n>Existing approaches typically rely on outcome-based rewards that are only provided at the final answer.<n>In this paper, we propose Information Gain-based Policy Optimization (IGPO), a simple yet effective RL framework that provides dense and intrinsic supervision for multi-turn agent training.
  arXiv  Detail & Related papers  (2025-10-16T17:59:32Z)
• Confidence as a Reward: Transforming LLMs into Reward Models [54.98336080630691]
  Confidence-as-a-Reward (CRew) is a training-free method that utilizes token-level confidence in the model's final answers as a proxy for reward.<n>We show that CRew outperforms existing training-free reward approaches on the MATH500 and RewardMATH benchmarks.<n>We propose CRew-DPO, a training strategy that constructs preference data from confidence scores combined with correctness signals.
  arXiv  Detail & Related papers  (2025-10-15T12:51:47Z)
• Agentic Reinforcement Learning with Implicit Step Rewards [92.26560379363492]
  Large language models (LLMs) are increasingly developed as autonomous agents using reinforcement learning (agentic RL)<n>We introduce implicit step rewards for agentic RL (iStar), a general credit-assignment strategy that integrates seamlessly with standard RL algorithms.<n>We evaluate our method on three challenging agent benchmarks, including WebShop and VisualSokoban, as well as open-ended social interactions with unverifiable rewards in SOTOPIA.
  arXiv  Detail & Related papers  (2025-09-23T16:15:42Z)
• LLM-Driven Intrinsic Motivation for Sparse Reward Reinforcement Learning [0.27528170226206433]
  This paper explores the combination of two intrinsic motivation strategies to improve the efficiency of learning agents in environments with extreme sparse rewards.<n>We propose integrating Variational State as Intrinsic Reward (VSIMR), which uses Variational AutoEncoders (VAEs) reward state novelty, with an intrinsic reward approach derived from Large Language Models (LLMs)<n>Our empirical results show that this combined strategy significantly increases agent performance and efficiency compared to using each strategy individually.
  arXiv  Detail & Related papers  (2025-08-25T19:10:58Z)
• Intra-Trajectory Consistency for Reward Modeling [67.84522106537274]
  We develop an intra-trajectory consistency regularization to enforce that adjacent processes with higher next-token generation probability maintain more consistent rewards.<n>We show that the reward model trained with the proposed regularization induces better DPO-aligned policies and achieves better best-of-N (BON) inference-time verification results.
  arXiv  Detail & Related papers  (2025-06-10T12:59:14Z)
• Reinforcing Multi-Turn Reasoning in LLM Agents via Turn-Level Reward Design [35.544075583073685]
  We present the first systematic study of textitturn-level reward design for multi-turn RL algorithms and agent applications.<n>We conduct case studies on multi-turn reasoning-augmented search agents, where we carefully design two types of turn-level rewards: verifiable and LLM-as-judge.<n>Our experiments on multi-turn search tasks demonstrate that incorporating well-designed turn-level rewards enables RL algorithms to significantly outperform baseline methods with trajectory-level rewards.
  arXiv  Detail & Related papers  (2025-05-17T04:09:46Z)
• From Novice to Expert: LLM Agent Policy Optimization via Step-wise Reinforcement Learning [62.54484062185869]
  We introduce StepAgent, which utilizes step-wise reward to optimize the agent's reinforcement learning process.<n>We propose implicit-reward and inverse reinforcement learning techniques to facilitate agent reflection and policy adjustment.
  arXiv  Detail & Related papers  (2024-11-06T10:35:11Z)
• Distributional Reward Estimation for Effective Multi-Agent Deep Reinforcement Learning [19.788336796981685]
  We propose a novel Distributional Reward Estimation framework for effective Multi-Agent Reinforcement Learning (DRE-MARL) Our main idea is to design the multi-action-branch reward estimation and policy-weighted reward aggregation for stabilized training. The superiority of the DRE-MARL is demonstrated using benchmark multi-agent scenarios, compared with the SOTA baselines in terms of both effectiveness and robustness.
  arXiv  Detail & Related papers  (2022-10-14T08:31:45Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.