Verified Critical Step Optimization for LLM Agents

• URL: http://arxiv.org/abs/2602.03412v1
• Date: Tue, 03 Feb 2026 11:41:02 GMT
• Title: Verified Critical Step Optimization for LLM Agents
• Authors: Mukai Li, Qingcheng Zeng, Tianqing Fang, Zhenwen Liang, Linfeng Song, Qi Liu, Haitao Mi, Dong Yu,
• Abstract summary: Critical Step Optimization focuses preference learning on verified critical steps.<n>Method starts from failed policy trajectories rather than expert demonstrations.<n>Experiments on GAIA-Text-103 and XBench-DeepSearch show that CSO achieves 37% and 26% relative improvement over the SFT baseline.
• Score: 67.05296684575445
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: As large language model agents tackle increasingly complex long-horizon tasks, effective post-training becomes critical. Prior work faces fundamental challenges: outcome-only rewards fail to precisely attribute credit to intermediate steps, estimated step-level rewards introduce systematic noise, and Monte Carlo sampling approaches for step reward estimation incur prohibitive computational cost. Inspired by findings that only a small fraction of high-entropy tokens drive effective RL for reasoning, we propose Critical Step Optimization (CSO), which focuses preference learning on verified critical steps, decision points where alternate actions demonstrably flip task outcomes from failure to success. Crucially, our method starts from failed policy trajectories rather than expert demonstrations, directly targeting the policy model's weaknesses. We use a process reward model (PRM) to identify candidate critical steps, leverage expert models to propose high-quality alternatives, then continue execution from these alternatives using the policy model itself until task completion. Only alternatives that the policy successfully executes to correct outcomes are verified and used as DPO training data, ensuring both quality and policy reachability. This yields fine-grained, verifiable supervision at critical decisions while avoiding trajectory-level coarseness and step-level noise. Experiments on GAIA-Text-103 and XBench-DeepSearch show that CSO achieves 37% and 26% relative improvement over the SFT baseline and substantially outperforms other post-training methods, while requiring supervision at only 16% of trajectory steps. This demonstrates the effectiveness of selective verification-based learning for agent post-training.

Related papers

• ProRAG: Process-Supervised Reinforcement Learning for Retrieval-Augmented Generation [54.071574153853994]
  ProRAG is a process-supervised reinforcement learning framework designed to integrate learned step-level supervision into the online optimization loop.<n>Our framework consists of four stages: (1) Supervised Policy Warmup to initialize the model with a structured reasoning format; (2) construction of an MCTS-based Process Reward Model (PRM) to quantify intermediate reasoning quality; (3) PRM-Guided Reasoning Refinement to align the policy with fine-grained process preferences; and (4) Process-Supervised Reinforcement Learning with a dual-granularity advantage mechanism.
  arXiv  Detail & Related papers  (2026-01-29T16:04:59Z)
• STEP: Success-Rate-Aware Trajectory-Efficient Policy Optimization [23.48518286261969]
  Trajectory-level optimization treats each trajectory as a single training sample.<n>This approach can be inefficient and yield misleading learning signals.<n>We propose STEP (Success-rate-aware Trajectory-Efficient Policy optimization), a framework that dynamically allocates sampling based on per-task success rates.
  arXiv  Detail & Related papers  (2025-11-17T07:43:15Z)
• Hybrid Reward Normalization for Process-supervised Non-verifiable Agentic Tasks [12.31210445905605]
  We introduce Principle Process Reward (PPR), an RL approach that unifies step-level assessment and outcome verification.<n>PPR achieves state-of-the-art performance across a wide range of benchmarks, demonstrating its impressive robustness and generalization.
  arXiv  Detail & Related papers  (2025-09-29T23:44:55Z)
• 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)
• Fast Adaptation with Behavioral Foundation Models [82.34700481726951]
  Unsupervised zero-shot reinforcement learning has emerged as a powerful paradigm for pretraining behavioral foundation models.<n>Despite promising results, zero-shot policies are often suboptimal due to errors induced by the unsupervised training process.<n>We propose fast adaptation strategies that search in the low-dimensional task-embedding space of the pre-trained BFM to rapidly improve the performance of its zero-shot policies.
  arXiv  Detail & Related papers  (2025-04-10T16:14:17Z)
• Look Before Leap: Look-Ahead Planning with Uncertainty in Reinforcement Learning [4.902161835372679]
  We propose a novel framework for uncertainty-aware policy optimization with model-based exploratory planning.<n>In the policy optimization phase, we leverage an uncertainty-driven exploratory policy to actively collect diverse training samples.<n>Our approach offers flexibility and applicability to tasks with varying state/action spaces and reward structures.
  arXiv  Detail & Related papers  (2025-03-26T01:07:35Z)
• Model Predictive Task Sampling for Efficient and Robust Adaptation [57.414812940406996]
  We introduce Model Predictive Task Sampling (MPTS), a framework that bridges the task space and adaptation risk distributions.<n>MPTS employs a generative model to characterize the episodic optimization process and predicts task-specific adaptation risk via posterior inference.<n>MPTS seamlessly integrates into zero-shot, few-shot, and supervised finetuning settings.
  arXiv  Detail & Related papers  (2025-01-19T13:14:53Z)
• Entropy-Regularized Process Reward Model [43.09203393852343]
  Large language models (LLMs) have shown promise in performing complex multi-step reasoning, yet they continue to struggle with mathematical reasoning.<n>We propose an entropy-regularized process reward model (ER-PRM) that integrates KL-regularized Markov Decision Processes (MDP)<n>Our empirical experiments on the MATH and GSM8K benchmarks demonstrate that ER-PRM consistently outperforms existing process reward models.
  arXiv  Detail & Related papers  (2024-12-15T01:09:23Z)
• 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)
• Monte Carlo Tree Search Boosts Reasoning via Iterative Preference Learning [55.96599486604344]
  We introduce an approach aimed at enhancing the reasoning capabilities of Large Language Models (LLMs) through an iterative preference learning process. We use Monte Carlo Tree Search (MCTS) to iteratively collect preference data, utilizing its look-ahead ability to break down instance-level rewards into more granular step-level signals. The proposed algorithm employs Direct Preference Optimization (DPO) to update the LLM policy using this newly generated step-level preference data.
  arXiv  Detail & Related papers  (2024-05-01T11:10:24Z)

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.