The code in this repository has been moved to https://github.com/hikvision-research/DoP.
Code of the CIKM 2025 paper "Decoder-only Pre-training Enhancement for Spatio-temporal Traffic Forecasting".