You can deploy this model and start earning money today!

mit-dailab/
watts

Views

No views yet

Watts: Pretrained Models for Synthetic Energy Time Series Generation

This repo contains pretrained checkpoints from the Watts model series, developed using the Cents library and trained on the PecanStreet dataset.

Watts models are designed to generate household-level electricity time series data—either for consumption only or combined with generation.

Model Name	Model Backbone	Description
Watts_1_1D	ACGAN	Fast generation of 1D synthetic consumption data via single forward pass.
Watts_1_2D	ACGAN	Fast generation of 2D synthetic consumption and generation data via single forward pass.
Watts_2_1D	Diffusion-TS	Slower, diffusion-based generation of higher fidelity 1D consumption data.
Watts_2_2D	Diffusion-TS	Slower, diffusion-based generation of higher fidelity 2D consumption + generation data.

MIT License