Robust Soft Skill Adaptation via Behavioral Localization in Tool-Using Agents· 工具使用代理的软技能适应
Text skills can extend the capabilities of tool-using agents, but long skill descriptions increase computational costs. Soft prompts provide a compact alternative, yet learning from entire successful trajectories may also introduce unnecessary behavioral changes. We find that the useful effect of a text skill is highly concentrated in a small subset of trajectory positions. Based on this observation, we propose Behavioral Localization, which compares the frozen agent with and without the text skill to identify positions where the skill both increases the likelihood of successful tokens and meaningfully changes the next token distribution. We use these localized positions to supervise a short soft prompt. Experiments on SpreadsheetBench with a frozen Qwen3.6-35B-A3B agent show that the top
通过行为定位优化工具使用代理的软技能适应。
- 核心方法
- 提出行为定位方法，通过比较有无文本技能的冻结代理，识别技能对成功令牌概率和下一令牌分布有显著影响的轨迹位置，用于监督短软提示。
- 适合谁读
- 研究者、工程师
- 要解决的问题
- 长文本技能描述增加了工具使用代理的计算成本，而从整个成功轨迹学习可能引入不必要的行为变化。
- 关键实验
- 在SpreadsheetBench数据集上使用Qwen3.6-35B-A3B冻结代理进行实验，表明方法有效。
- 主要贡献
- 减少了计算成本，提高了软技能的适应性和针对性。
- 意义与局限
- 该方法有助于提高工具使用代理的效率和准确性，但可能局限于特定类型的任务和数据集。