search

cvlab-stonybrook / PathLDM

Official Code for PathLDM: Text conditioned Latent Diffusion Model for Histopathology (WACV 2024)
33 stars 3 forks source link

VAE training questions #26

Closed bhosalems closed 3 days ago

bhosalems commented 2 weeks ago

How long was VAE trained? Number of steps or epochs? Also, was it also trained on the crops or the original image sizes?

srikarym commented 3 days ago

It was trained for three epochs on 256x256 crops (same training dataset as the LDM).