improved-diffusion
improved-diffusion copied to clipboard
openai
Release for Improved Denoising Diffusion Probabilistic Models
Has anyone made the code work for 1D images? I changed in_channels and out_channels to 1 in the creation of the model in script_util.py and removed the conversion to RGB...
Do I need to take any other steps during the training process? After training for a few days, the loss did not continue to decrease and the sampling results obtained...
Just wondering where is the training datasets，the link is unavailable.
I found this part quite confusing in train_util.py ``` if last_batch or not self.use_ddp: losses = compute_losses() else: with self.ddp_model.no_sync(): losses = compute_losses() ``` Is this part supposed to be...
It's an error when I run image_train.py My torch version is 1.7.1+cu110 Traceback (most recent call last): File "scripts/image_train.py", line 83, in main() File "scripts/image_train.py", line 41, in main TrainLoop(...
I can only generate 64x64 images with a 128x128 dataset, adjusting the image_size in script_util to 128 will show 'unsupported image size', adjusting it to 256 will show ' Error(s)...
`python scripts/image_sample.py --model_path /path/to/model.pt $MODEL_FLAGS $DIFFUSION_FLAGS` How come when I run this combination the sampled images are just static noises? I get that the model is unconditional, is that why?...
As in the title, I am wandering if I need to pass the class label when sampling a trained super resolution network. Isn't it wrong to use the data groundtruth...
Thanks for contributing such a great work. Why use 'zero_module' in the last convolution layer?
