arXiv cs.AI· Hariharan Devarajan, Neeraj Rajesh, Aniruddha Marathe, Nikoli Dryden, Kin Ng, Michela Taufer·· 2 小时前AI 评分50
Phronix：面向智能体的以能力为中心的学习范式
Phronix: The Capability-Centric Learning Paradigm for Agents
AI 导读
论文提出 Phronix，一种以能力为中心的学习范式，让基于大语言模型的智能体 harness 从自身执行中获取、验证并沉淀可复用能力，通过 predictor-validator-corrector 循环把交互提炼为能力库。
正文 · 原文
Abstract:Scientific computing increasingly relies on large language model-based agents for performance engineering tasks such as instrumentation, analysis, and workflow porting. However, agent effectiveness depends heavily on the surrounding harness, which typically relies on manually written skills and curated memory. This limits autonomous learning, forcing agents to repeatedly rediscover solutions rather than retain and reuse knowledge from prior executions. We introduce Phronix, a capability-centric learning paradigm that enables agentic harnesses to acquire, validate, and refine reusable capabilities from their own executions. Phronix operates through a predictor-validator-corrector loop: MCP tools and associated skills propose solutions, tool executions validate them, and observed outcomes refine the stored capabilities. It distills every observed interaction into a capability store and researches a failing task rather than abandoning it. First, such learned capabilities solve a recurring task on the first attempt in 93-95% of cases across every model, significantly reducing solution non-determinism and improving reliability. Second, Phronix turns an agent's own execution into 7.6-30.8 durable capabilities per task without user instruction, and creation falls below two once the needed capability exists. Third, warm-start runs consume only 0.38-0.40 million tokens per task, which is three to four times less than a single baseline attempt, while a cold-start costs no more than the baseline. On the 1000 Genome Pegasus workflow, instrumenting a production application for I/O performance analysis, Phronix passes all 11 correctness checks against six for the baseline harness, and also lifts the task solving rates of two cheaper models from 29% and 27% of SWE-bench tasks solved to 91% and 87%, respectively, matching the performance of the most expensive evaluated model.
| Comments: | 11 pages plus references |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.31789 [cs.AI] |
| (or arXiv:2610.31789v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.31789 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hariharan Devarajan [view email]
[v1]
Thu, 22 Oct 2026 18:29:52 UTC (204 KB)
来源：arXiv cs.AI · arxiv.org