Skip to content

npclu0609/STformer-and-NSTformer

 
 

Repository files navigation

STformer-and-NSTformer

This program contains the code and uesd datasets by STformer and NSTformer.

How to Run

Download

    cd /path/to/your/project
    git clone 

Important Notice

For using the two datasets METR-LA and PEMS-BAY, please unzip the file named “data_in_12_out_12_rescale_True.zip” in the two dataset dictionary first.

Run it

You can reproduce these models by running the following command:

    python experiments/train.py -c baselines/${MODEL_NAME}/${DATASET_NAME}.py --gpus '0'

Replace ${DATASET_NAME} and ${MODEL_NAME} with any supported models and datasets. For example, you can run NSTformer on METR-LA dataset by:

  python experiments/train.py -c baselines/NSTformer/METR-LA.py --gpus '0'

Multiple GPUs

To run model on multiple GPUs, just modify "CFG.GPU_NUM" in dataset named file under model name, such as ".../NSTformer/METR-LA.py" then modify the gpu nums in above command, for example:

  python experiments/train.py -c baselines/NSTformer/METR-LA.py --gpus '0,1,2,3'

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 57.1%
  • Python 42.5%
  • Shell 0.4%