flowers-dit-35m
A 35.5M-parameter diffusion transformer (Dew's SimpleDiT, patch size 4, 12 layers, width 384) trained
with EDM on Oxford Flowers at 128x128 for 30,000 steps, batch 64, by
Dew's tutorial 02.
This is that run's directory, with its final checkpoint; it holds the averaged (EMA) weights too.
import dew
task = dew.pipeline("dewml/flowers-dit-35m")
Tutorial 04 compares samplers on it.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support