Mingbird:面向小开源模型的本地优先 Agent harness 论文发布 v2
Mingbird: A Local-First Agent Harness Enabling Small Open Models to Complete Real Tasks
AI 导读
Hao Wang 与 Ting Huang 发布开源 harness Mingbird(arXiv:2610.02001v2),面向 1B-9B 参数小开源模型,可在消费级显卡上运行,包含 12k 行 Python、628 项测试和覆盖上下文经济、完成度、时间预算与宿主边界的 18 个机制。
正文 · 原文
Abstract:Small open-weight language models now run on a laptop, but they fail at multi-step work in ways a large model does not: they stop with the files missing, repeat one action until the budget ends, and report success without checking. At the one-to-nine-billion-parameter tier the harness supplies almost all of the procedural control that a large model supplies for this http URL present Mingbird, an open-source harness for that tier on a consumer graphics card: 12k lines of Python, a 628-test suite, and eighteen mechanisms for context economy, completion, time budget and the host boundary, plus a crawl guard outside those four groups. Its design point is completion control: the harness runs executable checks on the working directory and refuses a claim of completion when they this http URL evaluations accompany it. A configuration-surface audit finds four defects that the 625-test suite of v3.1.1 missed, each because a test checked a property adjacent to the one that broke. A ten-task probe suite gives each mechanism a reason to fire and reports where the change reaches the artifact, including one mechanism that cuts the signal it protects. Disabling five of six completion-time checks costs 0.152 of the artifact score on eighteen held-out tasks, three replications per cell (95% CI [+0.064, +0.246], Holm p = 0.014). On a 288-cell matrix the harness reaches 0.886 against 0.631, 0.479 and 0.405 for three open scaffolds, and on {\tau} 2 -bench it ranks first or tied first in all three domains against the arms this http URL, scorer, source and every per-cell result are public.
| Comments: | 67 pages, 30 tables, 4 figures. v2: substantially revised (18 mechanisms, the 288-cell v1.5.0 matrix, five evaluations). Code, benchmark protocol, scoring code and all per-cell results: this https URL |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.02001 [cs.AI] |
| (or arXiv:2610.02001v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02001 arXiv-issued DOI via DataCite |
Submission history
From: Hao Wang [view email]
[v1]
Thu, 1 Oct 2026 16:32:07 UTC (150 KB)
[v2]
Tue, 17 Nov 2026 18:43:13 UTC (107 KB)
来源:arXiv cs.AI · arxiv.org