Evidence Over Plans: Online Trajectory Verification for Skill Distillation

• URL: http://arxiv.org/abs/2605.09192v1
• Date: Sat, 09 May 2026 22:15:13 GMT
• Title: Evidence Over Plans: Online Trajectory Verification for Skill Distillation
• Abstract summary: Posterior Distillation Index (PDI) is a trajectory-level metric that quantifies how well a distilled skill is grounded in the task-environment evidence.<n> SPARK generates environment-verified trajectories used to compute PDI.<n>We show that SPARK-generated skills consistently surpass no-skill baselines and outperform human-written skills on student models.
• Score: 59.17625804241461
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Agent skills can remarkably improve task success rates by using human-written procedural documents, but their quality is difficult to assess without environment-grounded verification. Existing skill generation methods heavily rely on preference logs rather than direct environment interaction, often yielding negligible or even degraded gains. We identify that it is a fundamental timing bottleneck: robust skills should be posterior-based, distilled from empirical environment interaction rather than prior plans. In this study, we introduce the Posterior Distillation Index (PDI), a trajectory-level metric that quantifies how well a distilled skill is grounded in the task-environment evidence. To operationalize PDI, we present SPARK (Structured Pipelines for Autonomous Runnable tasKs and sKill generation) for preserving task execution evidence towards full trajectory-level analysis. SPARK generates environment-verified trajectories used to compute PDI, and it applies PDI as an online diagnostic and intervention signal to ensure posterior skill formation. Across 86 runnable tasks, SPARK-generated skills consistently surpass no-skill baselines and outperform human-written skills on student models (inference cost up to 1,000x cheaper than teacher models). These findings show that PDI-guided distillation produces efficient and transferable skills grounded in the task-environment interaction. We release our code at https://github.com/EtaYang10th/spark-skills .

Related papers

• Evo-Harness: Context-to-Harness Skill Compilation for Self-Evolving Agents [67.97365535829098]
  Learning from experience is critical for developing capable, self-improving large language model (LLM) agents.<n>We propose Evo-Harness, which distills noisy, single-shot executions into reusable skill harnesses for cross-domain and topic-level adaptation.<n>Our analysis demonstrates the effectiveness of Evo-Harness and provides a principled understanding of how LLM agents can effectively learn on the fly.
  arXiv  Detail & Related papers  (2026-08-15T06:43:56Z)
• OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning [38.54413500261524]
  On-policy self-distillation offers dense token-level supervision.<n>textbfOPID (textbfOn-textbfPolicy Sktextbfill textbfDistillation) is a framework that extracts skill supervision directly from completed on trajectories.
  arXiv  Detail & Related papers  (2026-06-25T09:24:09Z)
• SkillAudit: Ground-Truth-Free Skill Evolution via Paired Trajectory Auditing [81.51044612408793]
  SkillAudit is a framework for evolving agent skills without ground-truth feedback.<n>To turn behavioral differences into edit guidance, SkillAudit uses Process-Aligned Contrastive Evaluation.<n>It routes edits through two pipelines: Refine removes noisy or irrelevant guidance from broadly useful skills, while Repair replaces passages that conflict with the task.
  arXiv  Detail & Related papers  (2026-06-12T08:20:09Z)
• SkillEvolBench: Benchmarking the Evolution from Episodic Experience to Procedural Skills [31.23929961213889]
  Large language model (LLM) agents accumulate rich episodic trajectories while solving real-world tasks.<n>We introduce SkillEvolBench, a diagnostic benchmark for evaluating this step from experience reuse to skill formation.<n>We find that current agents often adapt locally but rarely form robust reusable skills.
  arXiv  Detail & Related papers  (2026-05-22T18:23:31Z)
• SkillGenBench: Benchmarking Skill Generation Pipelines for LLM Agents [15.669318649782285]
  We introduce SkillGenBench, a benchmark for evaluating skill generation pipelines under a unified protocol.<n>A generator receives raw corpora and produces standardized skill artifacts, which are then executed under fixed harnesses.<n> Experiments across a range of skill-generation methods and backbones show substantial performance variation.
  arXiv  Detail & Related papers  (2026-05-18T17:28:36Z)
• E$^2$DT: Efficient and Effective Decision Transformer with Experience-Aware Sampling for Robotic Manipulation [12.326967455610536]
  Decision Transformer (DT) has emerged as an effective framework for addressing long-horizon tasks.<n>E$2$DT is a DT-guided k-Determinantal Point Process sampling framework.<n>Our framework is experience-aware, allowing E$2$DT to be both efficient.
  arXiv  Detail & Related papers  (2026-04-30T19:28:44Z)
• Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents [24.99615788156812]
  We introduce Skill-SD, a framework that turns the agent's own trajectories into dynamic training-only supervision.<n>We derive an importance-weighted reverse-KL loss to provide gradient-correct token-level distillation.<n> Experimental results on agentic benchmarks demonstrate that Skill-SD substantially outperforms the standard RL baseline.
  arXiv  Detail & Related papers  (2026-04-12T14:57:52Z)
• Autonomous Continual Learning of Computer-Use Agents for Environment Adaptation [57.65688895630163]
  We introduce ACuRL, an Autonomous Curriculum Reinforcement Learning framework that continually adapts agents to specific environments with zero human data.<n>Our method effectively enables both intra-environment and cross-environment continual learning, yielding 4-22% performance gains without forgetting existing environments.
  arXiv  Detail & Related papers  (2026-02-10T23:06:02Z)
• Self-Imitated Diffusion Policy for Efficient and Robust Visual Navigation [7.341858898582114]
  Self-Imitated Diffusion Policy (SIDP) is a novel framework that learns improved planning by selectively imitating a set of trajectories sampled from itself.<n>Specifically, SIDP introduces a reward-guided self-imitation mechanism that encourages the policy to consistently produce high-quality trajectories efficiently.
  arXiv  Detail & Related papers  (2026-01-30T13:27:59Z)
• No Regrets: Investigating and Improving Regret Approximations for Curriculum Discovery [53.08822154199948]
  Unsupervised Environment Design (UED) methods have gained recent attention as their adaptive curricula promise to enable agents to be robust to in- and out-of-distribution tasks. This work investigates how existing UED methods select training environments, focusing on task prioritisation metrics. We develop a method that directly trains on scenarios with high learnability.
  arXiv  Detail & Related papers  (2024-08-27T14:31:54Z)
• Skill-Based Reinforcement Learning with Intrinsic Reward Matching [77.34726150561087]
  We present Intrinsic Reward Matching (IRM), which unifies task-agnostic skill pretraining and task-aware finetuning. IRM enables us to utilize pretrained skills far more effectively than previous skill selection methods.
  arXiv  Detail & Related papers  (2022-10-14T00:04:49Z)
• Design of Negative Sampling Strategies for Distantly Supervised Skill Extraction [19.43668931500507]
  We propose an end-to-end system for skill extraction, based on distant supervision through literal matching. We observe that using the ESCO taxonomy to select negative examples from related skills yields the biggest improvements. We release the benchmark dataset for research purposes to stimulate further research on the task.
  arXiv  Detail & Related papers  (2022-09-13T13:37:06Z)
• Learning Task Automata for Reinforcement Learning using Hidden Markov Models [37.69303106863453]
  This paper proposes a novel pipeline for learning non-Markovian task specifications as succinct finite-state task automata' We learn a product MDP, a model composed of the specification's automaton and the environment's MDP, by treating the product MDP as a partially observable MDP and using the well-known Baum-Welch algorithm for learning hidden Markov models. Our learnt task automaton enables the decomposition of a task into its constituent sub-tasks, which improves the rate at which an RL agent can later synthesise an optimal policy.
  arXiv  Detail & Related papers  (2022-08-25T02:58:23Z)
• Self-Knowledge Distillation with Progressive Refinement of Targets [1.1470070927586016]
  We propose a simple yet effective regularization method named progressive self-knowledge distillation (PS-KD) PS-KD progressively distills a model's own knowledge to soften hard targets during training. We show that PS-KD provides an effect of hard example mining by rescaling gradients according to difficulty in classifying examples.
  arXiv  Detail & Related papers  (2020-06-22T04:06:36Z)

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.