BUAABIGSCity / PDFormer

[AAAI2023] A PyTorch implementation of PDFormer: Propagation Delay-aware Dynamic Long-range Transformer for Traffic Flow Prediction.
MIT License
209 stars 35 forks source link

How to understand task level #33

Open suzy0223 opened 11 months ago

suzy0223 commented 11 months ago

task-level hasn't been mentioned in the paper. Would the task level influence the performance, why adding the task level?

aptx1231 commented 11 months ago

what is task level?

suzy0223 commented 10 months ago

I don't know. Task level appear when running the code.

suzy0223 commented 10 months ago

what is task level?

In PDFormer.py, use_curriculum_learning has been set as True and introduced the taks_level. In PDFormer easily define the task_level as time step needed to be predicted. But this haven't been mentioned in the paper.

aptx1231 commented 8 months ago

please refer to MTGNN model https://github.com/nnzhan/MTGNN