regarding the setting of the hyperparameter "ada_kimg"

[xuejuhao]

Hello, thank you for your excellent work. Regarding your code, there is one point that I am not clear about, what does the parameter "ada_kimg" represent?

if update_diffusion: # batch_idx % ada_interval == 0 adjust = np.sign(sign(Discriminator(real_images)) - ada_target) * C # C = (batch_size * ada_interval) / (ada_kimg * 1000) diffusion.p = (diffusion.p + adjust).clip(min=0., max=1.) diffusion.update_T()

[Zhendong-Wang]

It is a parameter to control the speed of adjusting the diffusion.p. If ada_kimg is large, diffusion.p will increase and decrease slowly. ada_kimg could also be seen as how many images need to be seen to let diffusion.p = 1 in the fastest increasing.

[xuejuhao]

Sure, thank you. May I know what values have you set for this parameter in your experiment? I would greatly appreciate it if you could provide me with an answer.

[Zhendong-Wang]

We use 100 and this is the default value in the code.

[xuejuhao]

We use 100 and this is the default value in the code.

fine, have a good day!