TARFVAE

May 16, 2025 ยท View on GitHub

Usage

  1. Install Pytorch and necessary dependencies.
pip install -r requirements.txt

  1. Due to space constraints, this work only includes ETTh1-related data. Additional open-source datasets referenced in our experiments can be found in the cited articles within the main text.

  2. Train and evaluate the model. We provide all the tasks under the folder ./scripts/.

# Multivariate forecasting with TARFVAE
sh ./scripts/multivariate_forecasting/ETT/TarfVAE_ETTh1.sh
  1. Using pre-trained checkpoints, we perform sampling with varying numbers of samples and calculate MSE, MAE, and CRPS metrics.
# Generate samples with TARFVAE
sh ./scripts/multivariate_forecasting/ETT/TarfVAE_ETTh1_crps.sh
  1. We have included the main experimental results logs in the ./logs/ folder