search

zoubohao / DenoisingDiffusionProbabilityModel-ddpm-

This may be the simplest implement of DDPM. You can directly run Main.py to train the UNet on CIFAR-10 dataset and see the amazing process of denoising.
MIT License
1.48k stars 156 forks source link

Can you upload some pretrained models for this project? #2

Closed mathfirst closed 2 years ago

mathfirst commented 2 years ago

I have no GPUs, so I would like to run with a trained model to save time. For example, you can upload the final checkpoints of Cifar10 code. I appreciate it if you can do that for me.

zoubohao commented 2 years ago

Thanks for asking this problem.

The link of diffusion weight is "https://drive.google.com/file/d/1B9jC_x72TxvcYWn8nOrUToIp10tPufSU/view?usp=sharing" The link of diffusion condition weight is "https://drive.google.com/file/d/1vtEbQEC_DZftDeuCgX8hg9a0oWCirji1/view?usp=sharing"

But you must notice that the diffusion inference is quite slow when only using the CPU. Good luck.