121644048/Transformer_Timeseries

Pytorch code for Google's Temporal Fusion Transformer

🐐 Time Series Forecasting

Time series forecasting models

Set-up

• Install the required packages

  • pip install -r requirements.txt

• Download data

  • https://drive.google.com/file/d/1Na7e2yJy1Oix8-HcKQS97u1VZodpZ-OZ/view?usp=sharing

• Train and Test on electricity dataset

  • python ./main.py --exp_name electricity --conf_file_path ./conf/electricity.yaml

  Plot prediction on Test set

  • python ./main.py --exp_name electricity --conf_file_path ./conf/electricity.yaml --inference=True

Models

• Temporal fusion transformer
  https://arxiv.org/pdf/1912.09363.pdf

  Usage:
  - model: tf_transformer

• Transformer
  https://arxiv.org/pdf/1706.03762.pdf
  https://pytorch.org/tutorials/beginner/transformer_tutorial.html

  Usage:
  - model: transformer

• GRN-Tranformer
  Use GRN block after multi-head attention to encode static variables

  Usage:
  - model: grn_transformer