metadata
license: openrail
This is a preliminary model based on the famous collection of images taken by Sergei Prokudin-Gorskii. They were taken between the years of 1909 and 1915 in pre-revolutionary Russia, and offer a glimpse into the past that few other datasets can claim.
- Based on sd 1.5 fork
- Dataset consists of 1,277 pre-composited images extracted from the Library of Congress public access portal
- 512px
- Trained on an A100 GPU for six hours at 2000 step epochs