Supplementary Material for ICLR 2023 Submission "Discrete
Predictor-Corrector Diffusion Models for Image Synthesis"
=========================================================

This ZIP file contains images with grids of 512x512 samples generated
by the following methods trained on ImageNet:

Ours:
- DPC-light(5) (Table 1)
- DPC-light(5) + Upsampling (Table 2)

Baselines:
- ADM+G+U (Table 2)
- StyleGAN-XL (Table 2)

The ImageNet class numbers and the method are indicated in each
filename. We included 8 randomly sampled images for each method and
for each class, for a subset of evenly spaced ImageNet classes.
