PATS: Policy-Aware Training Scaffolding for Agentic Reinforcement Learning

• URL: http://arxiv.org/abs/2607.21419v2
• Date: Thu, 30 Jul 2026 03:55:54 GMT
• Title: PATS: Policy-Aware Training Scaffolding for Agentic Reinforcement Learning
• Abstract summary: We propose a policy-centric training paradigm that reframes skills as a dynamic training scaffold.<n>Our framework, PATS, converts rollout groups from the latest policy into evidence cards and uses task-specific evaluation to adjust the context used in subsequent rollouts.
• Score: 8.197606162420145
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In long-horizon LLM agent reinforcement learning, weak policies often repeat similar failures, producing uninformative rollout trajectories and limiting effective policy optimization. Existing skill-centric methods improve exploration by optimizing, filtering, or internalizing reusable skills. However, they remain centered on the skills themselves rather than being designed as adaptive training-time support for the evolving policy. To address this, we propose a policy-centric training paradigm that reframes skills as a dynamic training scaffold. Our framework, PATS, converts rollout groups from the latest policy into evidence cards and uses task-specific evaluation to adjust the context used in subsequent rollouts. Concrete guidance helps weak policies to complete challenging tasks. As policy improves, redundant context is revised or removed to reduce reliance on explicit guidance while preserving useful rollout variation. The policy is optimized with environmental rewards using standard RLVR, and the training scaffold is discarded at deployment. Across ALFWorld, WebShop, and seven search-augmented QA benchmarks, PATS achieves performance competitive with SOTA baselines while using 25%-50% fewer tokens.

Related papers

• SPyCE: Skill-Policy Co-evolution for Multimodal Agents [12.105041985483167]
  We propose SPyCE, a framework that distills trajectories into a hierarchical skill library and updates it throughout reinforcement learning.<n>During training, the policy model conditions on retrieved skills to guide its rollouts, while the skill library evolves using valuable rollouts generated by the policy.<n> Experiments across eight benchmarks demonstrate that SPyCE consistently outperforms both RL-based and memory-based baselines.
  arXiv  Detail & Related papers  (2026-07-15T14:01:48Z)
• LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL [64.4276583943816]
  Reinforcement learning for non-verifiable instruction relies on judges with prompt-specific rubrics as reward signals.<n>While recent methods adapt these rubrics to the evolving policy during training, the training prompts themselves remain static.<n>This static approach often results in a critical misalignment between prompt difficulty and policy capability, leaving the judge unable to recover a discriminative reward signal.
  arXiv  Detail & Related papers  (2026-07-05T17:05:13Z)
• Beyond Action Residuals: Real-World Robot Policy Steering via Bottleneck Latent Reinforcement Learning [42.74142065376427]
  We propose Z-Perturbation Reinforcement Learning (ZPRL) to steer pretrained policies through a compact bottleneck latent.<n>In the real world, ZPRL improves the average success rate on four tasks by 33.7% over imitation base policies.
  arXiv  Detail & Related papers  (2026-05-19T14:43:26Z)
• Learning to Learn-at-Test-Time: Language Agents with Learnable Adaptation Policies [48.31565681022125]
  Test-Time Learning (TTL) enables language agents to iteratively refine their performance through repeated interactions with the environment at inference time.<n>We argue that optimal adaptation policies should be learned from task environments, not hand-engineered based on human intuition.<n>We introduce Meta-TTL, a framework that formulates the discovery of effective adaptation policies as a bi-level optimization problem.
  arXiv  Detail & Related papers  (2026-04-01T12:41:01Z)
• Learning in Context, Guided by Choice: A Reward-Free Paradigm for Reinforcement Learning with Transformers [55.33468902405567]
  We propose a new learning paradigm, In-Context Preference-based Reinforcement Learning (ICPRL), in which both pretraining and deployment rely solely on preference feedback.<n>ICPRL enables strong in-context generalization to unseen tasks, achieving performance comparable to ICRL methods trained with full reward supervision.
  arXiv  Detail & Related papers  (2026-02-09T03:42:16Z)
• A Step Back: Prefix Importance Ratio Stabilizes Policy Optimization [58.116300485427764]
  Reinforcement learning post-training can elicit reasoning behaviors in large language models.<n> token-level correction often leads to unstable training dynamics when the degree of off-policyness is large.<n>We propose a simple yet effective objective, Minimum Prefix Ratio (MinPRO)
  arXiv  Detail & Related papers  (2026-01-30T08:47:19Z)
• Stabilizing Policy Gradients for Sample-Efficient Reinforcement Learning in LLM Reasoning [77.92320830700797]
  Reinforcement Learning has played a central role in enabling reasoning capabilities of Large Language Models.<n>We propose a tractable computational framework that tracks and leverages curvature information during policy updates.<n>The algorithm, Curvature-Aware Policy Optimization (CAPO), identifies samples that contribute to unstable updates and masks them out.
  arXiv  Detail & Related papers  (2025-10-01T12:29:32Z)
• Relative Entropy Pathwise Policy Optimization [66.03329137921949]
  We present an on-policy algorithm that trains Q-value models purely from on-policy trajectories.<n>We show how to combine policies for exploration with constrained updates for stable training, and evaluate important architectural components that stabilize value function learning.
  arXiv  Detail & Related papers  (2025-07-15T06:24:07Z)
• On-Policy RL with Optimal Reward Baseline [109.47676554514193]
  On-Policy RL with Optimal reward baseline (OPO) is a novel and simplified reinforcement learning algorithm.<n>OPO emphasizes the importance of exact on-policy training, which empirically stabilizes the training process and enhances exploration.<n>Results demonstrate OPO's superior performance and training stability without additional models or regularization terms.
  arXiv  Detail & Related papers  (2025-05-29T15:58:04Z)
• FDPP: Fine-tune Diffusion Policy with Human Preference [57.44575105114056]
  Fine-tuning Diffusion Policy with Human Preference learns a reward function through preference-based learning.<n>This reward is then used to fine-tune the pre-trained policy with reinforcement learning.<n>Experiments demonstrate that FDPP effectively customizes policy behavior without compromising performance.
  arXiv  Detail & Related papers  (2025-01-14T17:15:27Z)
• Dynamic Learning Rate for Deep Reinforcement Learning: A Bandit Approach [2.743898388459522]
  In deep Reinforcement Learning (RL), the learning rate critically influences both stability and performance, yet its optimal value shifts during training as the environment and policy evolve.<n>Standard decay schedulers assume monotonic convergence and often misalign with these dynamics, leading to premature or delayed adjustments.<n>We introduce LRRL, a meta-learning approach that dynamically selects the learning rate based on policy performance rather than training steps.
  arXiv  Detail & Related papers  (2024-10-16T14:15:28Z)
• Skill-Critic: Refining Learned Skills for Hierarchical Reinforcement Learning [48.75878234995544]
  We propose the Skill-Critic algorithm to fine-tune the low-level policy in conjunction with high-level skill selection. We validate Skill-Critic in multiple sparse-reward environments, including a new sparse-reward autonomous racing task in Gran Turismo Sport.
  arXiv  Detail & Related papers  (2023-06-14T09:24:32Z)
• Simplifying Deep Reinforcement Learning via Self-Supervision [51.2400839966489]
  Self-Supervised Reinforcement Learning (SSRL) is a simple algorithm that optimize policies with purely supervised losses. We show that SSRL is surprisingly competitive to contemporary algorithms with more stable performance and less running time.
  arXiv  Detail & Related papers  (2021-06-10T06:29:59Z)

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.