← 研踪 PaperTrailarXiv:2611.24221 · v1
Learning to Coordinate: LLM-Guided Workflow Evolution for Token-Efficient Red Teaming
首次提交 2026-11-18 18:58
版本更新 2026-11-18 18:58
雷达观测 2026-11-19 15:04
摘要
Automated red teaming is essential for identifying safety failures in large language models (LLMs), but its practical scalability depends on the resources required to discover those failures. Multi-agent systems can combine complementary capabilities, yet repeated planning, redundant exchanges, and ineffective information routing can make their coordination expensive. Moreover, a target model's response indicates the outcome of a test without identifying which collaboration decision should change. We present Learning to Coordinate (L2C), an LLM-guided framework that evolves an executable multi-agent workflow for each red-teaming case. L2C represents agent responsibilities, execution dependencies, and input bindings in a versioned graph, and associates each graph with records of the information actually delivered and produced. A controller trained through supervised fine-tuning integrates these records with target responses, evaluator verdicts, and remaining budget to retain the workflow, refine instructions, restructure collaboration, or lock further editing. This design connects feedback to explicit changes in subsequent execution while allowing a useful organization to continue without repeated controller intervention. The evaluation protocol covers HarmBench with three execution backbones and two target models, together with component ablations, dataset transfer, and evaluator checks. Across the six executor-target configurations, L2C improves ASR over EvoSynth by 1.4 to 4.5 percentage points while reducing mean online tokens by 35.6% to 37.3%. L2C connects adaptive collaboration to the resource requirements of automated security evaluation.
作者备注:10 pages, 3 figures
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 18:58 · 最新观测版本
Learning to Coordinate: LLM-Guided Workflow Evolution for Token-Efficient Red Teaming
Automated red teaming is essential for identifying safety failures in large language models (LLMs), but its practical scalability depends on the resources required to discover those failures. Multi-agent systems can combine complementary capabilities, yet repeated planning, redundant exchanges, and ineffective information routing can make their coordination expensive. Moreover, a target model's response indicates the outcome of a test without identifying which collaboration decision should change. We present Learning to Coordinate (L2C), an LLM-guided framework that evolves an executable multi-agent workflow for each red-teaming case. L2C represents agent responsibilities, execution dependencies, and input bindings in a versioned graph, and associates each graph with records of the information actually delivered and produced. A controller trained through supervised fine-tuning integrates these records with target responses, evaluator verdicts, and remaining budget to retain the workflow, refine instructions, restructure collaboration, or lock further editing. This design connects feedback to explicit changes in subsequent execution while allowing a useful organization to continue without repeated controller intervention. The evaluation protocol covers HarmBench with three execution backbones and two target models, together with component ablations, dataset transfer, and evaluator checks. Across the six executor-target configurations, L2C improves ASR over EvoSynth by 1.4 to 4.5 percentage points while reducing mean online tokens by 35.6% to 37.3%. L2C connects adaptive collaboration to the resource requirements of automated security evaluation.
查看此版本解读 →查看此版本原文 ↗