The public repository of diffusion-based energy time series data generation. Paper: EnergyDiff: Universal Time-Series Energy Data Generation using Diffusion Models.
- Install the required dependency packages. This is the easiest to be done with
uv. Search for and install the package manageruvfirst. Then,
cd path/to/this/repository
uv sync-
Prepare the datasets by downloading from their original sources and running the preprocessing scripts (under
scripts/python/preprocessing). -
Training, sampling, and evaluation. Refer to the scripts (both for bash, for slurm, and for python) under
scripts/directory.
This repository adopts an MIT License. The conditional copula model is implemented by conditonal-copula .