斯坦福论文:Agent循环需改Harness,坏计划需微调权重
New Stanford paper finds that harness changes fix agents that loop or stall, whi…
Agent开发者的必读指南,清晰给出了Process vs Content故障的差异化优化路径,数据详实可直接指导工程实践。
New Stanford paper finds that harness changes fix agents that loop or stall, while agents that deliver bad plans need weight training instead.
斯坦福大学最新论文发现,调整智能体框架(harness)可以修复陷入循环或停滞的智能体,而提供糟糕计划的智能体则需要通过权重训练来改进。
An agent can be improved by editing its harness, the prompts, tools, and checks around the model, or by fine-tuning its weights.
可以通过编辑智能体框架——即模型周围的提示词、工具和检查机制——或者通过微调其权重来提升智能体的性能。
They sorted failed runs into process failures, such as loops and used-up step budgets, and content failures, where a poor plan was delivered. On a travel-planning benchmark, an LLM-driven loop rewrote the harness, and its best runs then fine-tuned the model.
他们将失败运行分为流程故障(如循环和步骤预算耗尽)和内容故障(即提供了糟糕的计划)。在旅行规划基准测试中,由大语言模型驱动的循环改进了智能体框架,其最佳运行结果随后用于对模型进行微调。
Harness evolution lifted Qwen3.5-4B from 0.16 to 0.30 on held-out tasks, as plan delivery rose from 55% to 90%. Harness edits never shrank the share of poor plans, but a LoRA adapter cut them from 28% to 5% of Qwen3.5-9B's held-out runs.
在保留任务上,框架优化使 Qwen3.5-4B 的性能从 0.16 提升至 0.30,计划交付率从 55% 上升至 90%。虽然框架编辑从未减少糟糕计划的比例,但一个 LoRA 适配器将 Qwen3.5-9B 在保留运行中的糟糕计划比例从 28% 降低到了 5%。
Before improving an agent, label why its runs fail, then fix process failures in the harness and content failures in the weights.
在改进智能体之前,应先标记其运行失败的原因,然后在框架中修复流程故障,在权重中修复内容故障。
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力