This is our first version of RR-Former in the following paper: RR-Former: Rainfall-runoff modeling based on Transformer. Thank you for your attention to our work. I'll release more elegant code in the future version.
- Using pretrain.py to pretrain a RR-Former. Furthermore, using pretrain_test_global.py to test the pretrained model in a global way, and using pretrain_test_single.py to test the pretrained model in a single-basin way.
- Using fine_tune.py to fine tune a RR-Former model (based on pretrained RR-Former). Furthermore, using fine_tune_test.py to test the fine-tuned model.
- Download data from the following paper: CAMELS: Catchment Attributes and MEteorology for Large-sample Studies.
- Set all configs in the directory configs.
- Change data-relevant configs in the directory data.