analiza aiqbench
SkillRise introduces a unified reinforcement learning framework that enables large language model agents to learn reusable skills across distinct tasks rather than treating them as independent episodes. By decoupling credit assignment between solving current problems and curating evolving skill documents, the method achieves significant performance gains on benchmarks like ALFWorld and WebShop while reducing runtime overhead compared to multi-stage pipelines. This approach demonstrates that LLMs can effectively transfer solution patterns across related tasks without relying solely on repeated sampling of identical instances.
SkillRise: Agentic Reinforcement Learning for Cross-Task Skill Evolution
arXiv:2607.26784v1 Announce Type: cross Abstract: Large language model agents often encounter related yet distinct tasks that share reusable solution patterns. Yet standard agentic reinforcement learning treats tasks as independent episodes, while existing approaches to skill learning either focus on repeated attempts of one task or use pipelines with multiple stages that entangle extraction, retrieval, and execution. We introduce SkillRise, a unified reinforcement learning framework for learning skills across tasks. SkillRise organizes related instances into progressively challenging sequences and uses a single policy to alternate between task solving and curating an evolving skill document passed directly to the next task. Decoupled credit assignment across tasks supervises solving with the current task outcome and curation with discounted downstream outcomes. Experiments on ALFWorld, WebShop, and ScienceWorld show that SkillRise achieves the strongest Pass@1 performance among the compared methods, with gains over the strongest baseline ranging from 2.3 to 8.5 percentage points. Although trained across distinct tasks, its learned curation policy remains effective for repeated attempts on the same task. Further analysis reveals scaling at test time across tasks: performance improves with longer sequences of related tasks even when each task is attempted only once. This trend suggests that SkillRise reuses transferable skills across tasks rather than benefiting from repeated sampling of the same task. SkillRise further retains strong performance while substantially reducing the runtime overhead of skill learning pipelines with multiple stages. Together, these results provide a simple and efficient training paradigm for LLM agents to extract, refine, and reuse transferable skills across tasks.
Abstract:Large language model agents often encounter related yet distinct tasks that share reusable solution patterns. Yet standard agentic reinforcement learning treats tasks as independent episodes, while existing approaches to skill learning either focus on repeated attempts of one task or use pipelines with multiple stages that entangle extraction, retrieval, and execution. We introduce SkillRise, a unified reinforcement learning framework for learning skills across tasks. SkillRise organizes related instances into progressively challenging sequences and uses a single policy to alternate between task solving and curating an evolving skill document passed directly to the next task. Decoupled credit assignment across tasks supervises solving with the current task outcome and curation with discounted downstream outcomes. Experiments on ALFWorld, WebShop, and ScienceWorld show that SkillRise achieves the strongest Pass@1 performance among the compared methods, with gains over the strongest baseline ranging from 2.3 to 8.5 percentage points. Although trained across distinct tasks, its learned curation policy remains effective for repeated attempts on the same task. Further analysis reveals scaling at test time across tasks: performance improves with longer sequences of related tasks even when each task is attempted only once. This trend suggests that SkillRise reuses transferable skills across tasks rather than benefiting from repeated sampling of the same task. SkillRise further retains strong performance while substantially reducing the runtime overhead of skill learning pipelines with multiple stages. Together, these results provide a simple and efficient training paradigm for LLM agents to extract, refine, and reuse transferable skills across tasks.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2607.26784 [cs.LG]
(or arXiv:2607.26784v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2607.26784 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 267 cuvinte
Subiecte conexe
The Alignment Problem: Computer Science Meets Analytic Philosophy
LessWrong · 18 oct. 2026
TTSE: A Two-Track Online Self-Evolution Framework
arXiv · 22 sept. 2026
Coalition-Aware Skill Reliability for Self-Evolving Agents
arXiv · 25 aug. 2026
ReSkill: Reconciling Skill Creation with Policy Optimization in Agentic RL
arXiv · 9 iun. 2026