大小模型协同优化的兴趣点轨迹预测框架

魏蕴田,金苍宏,费峥东,郑铜亚,王晓亮,宋明黎

Collaborative optimization framework of large and small model for POI trajectory prediction

Yuntian WEI,Canghong JIN,Zhengdong FEI,Tongya ZHENG,Xiaoliang WANG,Mingli SONG

表 7 纠错前、后的Acc@1对比

Tab.7 Comparison of Acc@1 before and after error correction

方法 TKY NYC CA 纠正前 纠正后 纠正前 纠正后 纠正前 纠正后 LLM-RFA(Deepseek-R1) 0.2743 0.3166 0.2848 0.3259 0.2727 0.3285 LLM-RFA(Qwen) 0.2523 0.3050 0.2474 0.2835 0.2609 0.2900 LLM-RFA(Gpt-4o) 0.2379 0.2800 0.2776 0.3162 0.2174 0.3125 LLM-RFA(Gpt-4) 0.2167 0.2381 0.1836 0.2222 0.2173 0.2692