search

leffff / adversarial-diffusion-distillation

My Implementation of Adversarial Diffusion Distillation https://arxiv.org/pdf/2311.17042.pdf
MIT License
33 stars 3 forks source link

Did you manage to get plausible results? #1

Open ascust opened 4 months ago

leffff commented 4 months ago

Hi, so far I am limited in compute, therefore I am waiting to soon get my hands on more powerful GPUs. Right now I only have RTX 4090, which is not enough

digbangbang commented 4 months ago

Hi, the paper says that the step size of the teacher model is set to 1. I think this is unreasonable. I tried to use ddpm CIFAR10 to conduct ADD experiments. When the teacher model step size is 1, sampling is performed, and the result is a picture of completely random noise.

leffff commented 4 months ago

answered here https://github.com/Stability-AI/generative-models/issues/238

Hi, the paper says that the step size of the teacher model is set to 1. I think this is unreasonable. I tried to use ddpm CIFAR10 to conduct ADD experiments. When the teacher model step size is 1, sampling is performed, and the result is a picture of completely random noise.