MAM-E weights
Collection
This collection contains the weights of several experiments of the MAM-E project
•
6 items
•
Updated
This is a latent diffusion model for generation of mammograms with size 64x64 in the latent space. It was trained with 4K images in the latent space (10% of the original dataset) for 50 epochs. Images in the latent space have size (4,64,64) and can generate images of 512x512 when decoded using the VAE.
from diffusers import DDPMPipeline
pipeline = DDPMPipeline.from_pretrained({hub_model_id})
image = pipeline().images[0]
image