arXiv cs.AI· Mohamed Abouzahra·· 2 天前AI 评分22
微调语言智能体的课程覆盖与规则特定失效：Learning When to Withdraw
Learning When to Withdraw: Curriculum Coverage and Rule-Specific Failures in Fine-Tuned Language Agents
AI 导读
研究在 NAQD-Env 合成环境中用 LoRA 微调 Qwen2.5-3B-Instruct，比较原始与增强课程对选择性撤回能力的影响。在 naqd 提示下，SFT 检查点的决策一致率达 0.77-0.92、撤回召回率 0.46-0.86，而三个基线仅为 0.45-0.63 和最高 0.06。
正文 · 原文
Abstract:Tool-using language agents must revise prepared actions when evidence changes or authority is revoked. Selective withdrawal means suspending affected actions, preserving supported work, and resuming only after sufficient repair. We study supervised fine-tuning in NAQD-Env, a synthetic environment with explicit dependencies and a deterministic reference policy. Four LoRA checkpoints of Qwen2.5-3B-Instruct compare an original curriculum with an enriched one, the latter under three system-prompt framings with identical views and targets. Three instruction-tuned baselines and all four checkpoints are evaluated under three inference prompts on fresh validation cases, held-out dependency families, and held-out structural combinations. Under the naqd prompt, SFT checkpoints reach decision agreement of 0.77-0.92 and withdrawal recall of 0.46-0.86, versus 0.45-0.63 and at most 0.06 for the baselines. Rule-level diagnostics qualify these gains. The original-curriculum checkpoint wrongly withdraws 26 of 34 actions with surviving alternative support; enriched checkpoints withdraw 0-1. Conversely, all SFT checkpoints narrow prohibited combinations less often than the 7B and 8B baselines, a case absent from both curricula. Delegation results show more withdrawal but poorer preservation of independently authorized actions, and all four checkpoints omit valid event reports. Framing contrasts vary across metrics and splits and do not establish a reproducible advantage or equivalence. One training run per arm, jointly changed curriculum properties, prior evaluation exposure, and baseline-SFT quantization differences limit causal attribution. The contribution is a method for detecting rule-specific regressions that aggregate accuracy can conceal, not evidence that fine-tuning alone ensures agent safety.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.21045 [cs.AI] |
| (or arXiv:2610.21045v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.21045 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mohamed Abouzahra [view email]
[v1]
Thu, 15 Oct 2026 00:38:39 UTC (853 KB)
来源：arXiv cs.AI · arxiv.org