A demo model from the Data & Impact Accounting (DIA) lab. It performs
denoising diffusion generative model (MNIST) via from-scratch DDPM, with the base model trained from scratch, trained on
NVIDIA A100.
The point of this repo is not the model itself but its dia_report — a
standardized record of the energy, carbon, and water used to train it, embedded
in this card's metadata.
This footprint feeds the DIA dashboard, which rolls up a base model and all its derivatives to show the cumulative carbon, water, and energy cost of a model family.
Training footprint
Metric
Value
Hardware
1× NVIDIA A100-SXM4-80GB
Compute
0.5017 GPU-hours
Energy
0.104 (measured) kWh
Carbon
0.0067 (measured) kgCO₂eq
Water
0.187–0.416 (estimated-from-default-wue) L
Grid region
ca-on
Energy and carbon are measured with CodeCarbon;
water is estimated from a default water-usage-effectiveness range. Carbon uses the
local grid's intensity (Ontario, ~0.03 kgCO₂eq/kWh).
