improved-diffusion
improved-diffusion copied to clipboard
openai
Release for Improved Denoising Diffusion Probabilistic Models
When I train the model for some steps,I often find that the loss value rise up to a very high level(such as 747),what happened?Is it related to the value of...
I have noticed that sampling with this model takes an extraordinary amount of time, far slower than any individual training loop. At 4000 diffusion steps, it seems to get through...
Hi, how should I modify the code to train the fashionmnist or mnist data set, their size is 28x28, which is not a supported size, how to modify the code...
I used the command you gave. The generated image is very dark. I trained for 10k rounds. Is this normal. Or you can give the record of loss value in...
------------------------ | grad_norm | 0.0913 | | loss | 0.0621 | | loss_q0 | 0.17 | | loss_q1 | 0.0455 | | loss_q2 | 0.0209 | | loss_q3 | 0.00557...
Hi, I am training on a dataset of 500 images (I know it's not a lot), with diffusion steps=4000 and basic parameters proposed in the read me. Using the first...
I am trying to train the improved DDPM without class-conditionals but setting the condition to False doesn't let me skip the following error. Am I missing something? File "/home/diffusion_experiments/improved_ddpm_experiments/main.py", line...
Hi! Thanks for your great work. I want to ask why it is difficult to converge on some custom datasets. These dataset have a very uniform hue and lighting (e.g.,...
``` python ~/improved-diffusion-main/scripts/image_train.py --data_dir ~/train1000 --image_size 64 --num_channels 128 --num_res_blocks 3 --diffusion_steps 4000 --noise_schedule linear --lr 1e-4 --batch_size 128 Logging to /tmp/openai-2023-02-08-04-54-58-637829 creating model and diffusion... creating data loader... training......
I'm trying to work with a custom model and due to a tech problem have to restart from a check point. When I use the --resume_checkpoint I get this loading...
