---
title: "工具使用代理跨语言动作策略保持度测量：行动胜于语言"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.11110v1"
permalink: "https://feed.astralor.com/items/itm_eb5c8af2ff93f784c590790b"
published: "2026-08-11T16:18:34.000Z"
score: 77
channel: "ai"
tags: ["arXiv","动作策略","工具使用Agent","跨语言评测"]
---

# 工具使用代理跨语言动作策略保持度测量：行动胜于语言

## AI 点评

> 这是今年少见的、把多语言评测的‘题眼’从答案挪到动作上的论文。对做 Agent 评估或部署多语言产品的团队，它提供了一面照妖镜：如果只根据文本输出判断模型能力，可能连 26 倍误差都发现不了。我尤其推荐关注那个‘非英语先经英语中转’的发现——它意味着当前模型的多语言能力本质上是‘英语策略 + 翻译通道’，这会影响成本、延迟和失败边界。

## 速览

这项 arXiv 新研究将工具型 AI Agent 的“动作策略”而非最终答案作为评测对象，在 8 个模型、6 个并行基准、41 种语言和 238 万次 rollout 中测量跨语言一致性，并消除了五个混淆因素。结果显示，前沿模型经自身可重复性归一化后，仅保留 71–73% 的动作策略，模型身份只解释 5.7% 的方差；小于约 10B 参数的模型则出现崩溃。研究证实非英语任务会经由英语“中转”，且无法用提示词消除；还发现一个简单的轨迹提取正则可伪造 26 倍准确率提升，提醒多语言评测必须审计动作而非只看输出。

## 关键点

- 首次以动作轨迹为评测对象，覆盖 8 个模型、41 种语言、238 万次 rollout，系统性识别并纠正五个混淆因素。
- 经可重复性归一化后，前沿模型跨语言动作保持率收敛至 71–73%，模型身份仅解释 5.7% 方差。
- 低于约 10B 参数的模型出现跨语言策略崩溃，小模型排序多数由偶然地板决定。
- 非英语任务经过英语中转是有因果支撑的行为，且无法通过提示词让模型放弃；提取正则可伪造 26 倍准确率提升。

## 全文

When a tool-using agent is given the same task in a different language, does it still take the same steps? Multilingual evaluation rarely asks: it compares final answers and discards the actions. Yet those actions are the product: they fix cost and latency, decide how the system fails, and are the only auditable part of its behaviour. We make the action policy the measured object across 8 models, 6 parallel benchmarks and 41 languages (2.38M rollouts). The naive measurement fails: five confounds sit between raw trace similarity and any defensible claim, each able to flip a conclusion. Short traces score higher, empty traces score perfectly, unrelated traces agree by chance over half the time, the gap is capped by each model's reproducibility, and a model asked the same question twice in one language answers differently, leaving no baseline. We remove all five, and every correction makes the effect larger. Divergence proves structural, not sampling noise: it survives greedy decoding in every cell and stays flat as temperature rises, even as models grow less self-consistent. Normalised by their own reproducibility, four very different frontier models converge under greedy decoding, each keeping 71-73% of its action policy across languages, with model identity explaining only 5.7% of the variance. Below roughly 10B parameters it breaks down, and the ordering among smaller models is largely an artifact of a chance floor we measure by permutation rather than assume. Agents route non-English tasks through English; this pivot is causally load-bearing, confirmed by a pre-registered prediction across four models, and models will not abandon it when told to. Finally, a single trace-extraction regex, not the model, manufactured a multilingual failure: two worked examples raise one model's measured accuracy twenty-sixfold while its accuracy on readable outputs barely moves.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_eb5c8af2ff93f784c590790b