A Dynamic Fusion Large Language Model for Traffic Flow Prediction
The model incorporates three core components: spatiotemporal embedding module, spatiotemporal fusion module, and LLM backbone, which adopts a differentiated parameter adaptation strategy to balance training efficiency and traffic data adaptability.
Xue Qiu, Jian-Li Xiao
· 0 citations