arXiv cs.CL· JunHyeok Oh, Byung-Jun Lee·· 12 小时前AI 评分38
StreamStop：为语言模型智能体实现流式推理的自适应停止
Adaptive Stopping of Streamed Reasoning for Language Model Agents
AI 导读
StreamStop 通过离线学习模型特定的停止策略，从前缀级动作一致性判断何时停止推理，无需访问推理模型内部信号。在 WebShop 和 ScienceWorld 闭环评测中，其成功率高于多数经验证集选择的固定预算设置，且推理 token 少于完整思考。在 WebShop 上用 Qwen3.8-27B 时，每回合推理 token 减少 42%，任务成功率相近。
正文 · 原文
Abstract:Reasoning-enabled language model agents can incur substantial computation before acting, although the reasoning needed varies across decisions. Existing stopping signals, such as confidence and intermediate-answer stability, can vary in effectiveness across reasoning models; signals obtained from hidden states or token probabilities also require internal access that may be unavailable or impractical. We introduce StreamStop, which learns model-specific stopping policies offline from prefix-level action agreement. A frozen prefix reader encodes available reasoning text for a lightweight stop head, allowing stopping without access to the reasoning model's internal signals. We construct and publicly release a dataset of agent reasoning trajectories with prefix-level action annotations. These annotations compare each prefix action with the full-reasoning reference action, enabling offline comparison of stopping rules and exact expected-return optimization of a discounted action-agreement objective. Conditioning the head on the discount factor allows the agreement-token trade-off to be adjusted at inference without retraining. In closed-loop evaluations on WebShop and ScienceWorld, StreamStop achieves higher observed success than most validation-selected fixed-budget settings while using fewer reasoning tokens than Full thinking. On WebShop with Qwen3.8-27B, it uses 42% fewer reasoning tokens per episode with similar observed task success.
| Comments: | 15 pages, 6 figures. Project page: this https URL |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.27981 [cs.CL] |
| (or arXiv:2610.27981v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.27981 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: JunHyeok Oh [view email]
[v1]
Tue, 20 Oct 2026 10:24:30 UTC (811 KB)
来源：arXiv cs.CL · arxiv.org