SPyCE: Skill-Policy Co-evolution for Multimodal Agents

• URL: http://arxiv.org/abs/2607.13854v1
• Date: Wed, 15 Jul 2026 14:01:48 GMT
• Title: SPyCE: Skill-Policy Co-evolution for Multimodal Agents
• Abstract summary: 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.
• Score: 12.105041985483167
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Multimodal agents that think with images iteratively manipulate visual evidence and invoke tools across many steps. Existing reinforcement learning methods reduce trajectories to scalar rewards, forcing the policy to discover reusable tool-use patterns from scratch on every new task; memory-based alternatives retain past experience, yet they rely on test-time retrieval, without updating the policy to absorb reusable patterns from that experience. Our key insight is that multimodal reasoning trajectories should be distilled into reusable skills that co-evolve with the policy during training, rather than being consumed as rewards or retrieved from a static store. To this end, we propose SPyCE (Skill-Policy Co-evolution), a framework that distills trajectories into a hierarchical skill library and updates it throughout reinforcement learning. Execution skills capture local visual operations, while workflow skills encode high-level priors that orchestrate tool use. 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. This creates a closed loop in which improved policies yield better skills, and the evolving skill library, in turn, provides stronger priors for policy rollouts. Experiments across eight benchmarks demonstrate that SPyCE consistently outperforms both RL-based and memory-based baselines. Further analysis reveals that both the hierarchical skill design and the co-evolution mechanism are critical to our design. These results suggest joint skill-policy optimization as a promising paradigm for building capable multimodal agents.

Related papers

• SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning [38.54413500261524]
  Outcome-based reinforcement learning (RL) provides a practical optimization paradigm, but its sparse trajectory-level rewards offer limited guidance on intermediate decisions.<n>We propose SEED (SElf-Evolving On-Policy Distillation), a self-evolving framework that converts completed on-policy trajectories into training-time hindsight skills.
  arXiv  Detail & Related papers  (2026-07-16T09:57:18Z)
• RL Forgets! Towards Continual Policy Optimization [57.4336338996653]
  Continual post-training is becoming a central paradigm for adapting vision-language models to evolving tasks.<n>Recent work has increasingly favored reinforcement learning over supervised fine-tuning, driven by the belief that reinforcement learning is inherently less prone to forgetting.<n>We propose Continual Policy Optimization (CPO), a replay-free framework grounded in a prior-task behavioral KL objective.
  arXiv  Detail & Related papers  (2026-07-05T15:46:11Z)
• ReSkill: Reconciling Skill Creation with Policy Optimization in Agentic RL [24.65745354725497]
  We introduce ReSkill, an RL-in-the-loop skill creation framework that reconciles skill evolution with policy learning.<n>ReSkill exploits the group-wise structure of GRPO to embed three mechanisms with only marginal additional overhead.<n>Across several domains, ReSkill consistently outperforms existing memory and skill-based RL methods.
  arXiv  Detail & Related papers  (2026-06-01T03:12:05Z)
• You Live More Than Once: Towards Hierarchical Skill Meta-Evolving [92.6633152185458]
  Test-time skill evolving is regarded as a new paradigm for enhancing agentic systems.<n>We show that test-time refinement of the skill evolving framework is necessary for continuous improvement of the agent systems.<n>We propose HiSME, a lightweight hierarchical skill meta-evolving solution.
  arXiv  Detail & Related papers  (2026-05-27T12:26:49Z)
• CODESKILL: Learning Self-Evolving Skills for Coding Agents [13.599750520978988]
  Coding agents produce rich trajectories while solving software-engineering tasks.<n>To enable agent self-evolution, these trajectories can be distilled into reusable procedural skills that compactly encode experience to guide future behavior.<n>We propose CODESKILL, an LLM-based framework that reformulates skill extraction and skill-bank maintenance as a learnable management policy.
  arXiv  Detail & Related papers  (2026-05-25T05:12:49Z)
• ARISE: Agent Reasoning with Intrinsic Skill Evolution in Hierarchical Reinforcement Learning [17.98540130851038]
  ARISE (Agent Reasoning via Intrinsic Skill Evolution) is a hierarchical reinforcement learning framework.<n>It uses a shared policy to manage skills at high-level and to generate responses at low-level.<n>A hierarchical reward design guides the co-evolution of reasoning ability and library quality.
  arXiv  Detail & Related papers  (2026-03-17T02:03:17Z)
• SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning [83.98129545309277]
  We propose SkillRL, a framework that bridges the gap between raw experience and policy improvement.<n>Our approach introduces an experience-based distillation mechanism to build a hierarchical skill library SkillBank.<n> Experimental results on ALF, WebShop and seven search-augmented tasks demonstrate that SkillRL achieves state-of-the-art performance.
  arXiv  Detail & Related papers  (2026-02-09T03:17:17Z)
• Scalable In-Context Q-Learning [68.9917436397079]
  We propose textbfScalable textbfIn-textbfContext textbfQ-textbfLearning (textbfSICQL) to steer in-context reinforcement learning.<n>textbfSICQL harnesses dynamic programming and world modeling to steer ICRL toward efficient reward and task generalization.
  arXiv  Detail & Related papers  (2025-06-02T04:21:56Z)
• Text Generation with Efficient (Soft) Q-Learning [91.47743595382758]
  Reinforcement learning (RL) offers a more flexible solution by allowing users to plug in arbitrary task metrics as reward. We introduce a new RL formulation for text generation from the soft Q-learning perspective. We apply the approach to a wide range of tasks, including learning from noisy/negative examples, adversarial attacks, and prompt generation.
  arXiv  Detail & Related papers  (2021-06-14T18:48:40Z)
• Meta-Reinforcement Learning Robust to Distributional Shift via Model Identification and Experience Relabeling [126.69933134648541]
  We present a meta-reinforcement learning algorithm that is both efficient and extrapolates well when faced with out-of-distribution tasks at test time. Our method is based on a simple insight: we recognize that dynamics models can be adapted efficiently and consistently with off-policy data.
  arXiv  Detail & Related papers  (2020-06-12T13:34:46Z)
• Efficient Deep Reinforcement Learning via Adaptive Policy Transfer [50.51637231309424]
  Policy Transfer Framework (PTF) is proposed to accelerate Reinforcement Learning (RL) Our framework learns when and which source policy is the best to reuse for the target policy and when to terminate it. Experimental results show it significantly accelerates the learning process and surpasses state-of-the-art policy transfer methods.
  arXiv  Detail & Related papers  (2020-02-19T07:30:57Z)

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.