Primary sourcearXiv: DeepSeek
View sourceEvent ·Official first-hand
Pezego-HITL:面向加纳农业扩展的策略导向大语言模型架构
本文提出P-EVAL评估框架,用于策略约束的大语言模型决策支持,在加纳模拟场查询数据库(1,240例)上测试。Pezego-HITL架构在P-EVAL下实现Policy Alignment Rate (PAR) 0.94,Agronomic Utility Rate (AUR) 0.95,P95延迟降低55%(从28.6s到12.9s)。通过59.6%缓存重用率实现延迟优化。在Qwen3.5-9B-DeepSeek-V4-Flash模型上泛化取得PAR 0.86,延迟降低54.5%(至10.2s)。对加纳30名推广官员和36名小农进行问卷评估,验证了实际效用。
Current read
这篇论文搞了个P-EVAL框架来评估农业AI的安全性,Pezego-HITL架构在加纳测试,延迟降了55%,还挺靠谱的。
2 sources41° AI heatLast updated 2026/7/15 15:14:49
Evidence
2 sourcesRelated source 1IT之家
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.