alexzhou907/ddbm

ddbm's Issues

How to run our own dataset

Thanks for your great work, I want to konw if the work could been used with unpaired dataset ,and if I use my own dataset, is there anything I need to change like load_data() in DDBM-main\ datassets _init_.py

logsnr function

Hi @alexzhou907 , It is a really nice work!
I've been going through the code in Line91 of karras_diffusion.py and Line93 of karras_diffusion.py of the same file, and I've got a question regarding the parameter 't' in the functions vp_logsnr and vp_logs. It seems to consistently have a value of 1. Could you kindly clarify why this value is chosen for 't'?

def vp_logsnr(t, beta_d, beta_min):
    t = th.as_tensor(t)
    return - th.log((0.5 * beta_d * (t ** 2) + beta_min * t).exp() - 1)
    
def vp_logs(t, beta_d, beta_min):
    t = th.as_tensor(t)
    return -0.25 * t ** 2 * (beta_d) - 0.5 * t * beta_min

code realse

Great work. When do you plan to release the code?

About VP Schedule's Hyperparamter

Hi @alexzhou907 ! It's really a nice work. It's thrilling to see it gain recognition and acceptance.

Upon reviewing both the paper and the source code, I've encountered a point of confusion regarding the implementation of the VP schedule's hyperparameter, specifically the constant drift term. The paper mentions that incorporating a constant drift term could enhance performance, which caught my attention.

In the code, however, it seems that there is no such implementation, or I might have overlook it. The most relevant section I found about VP schedule implementation is this line of code, which seems to still use the unbalanced diffusion bridge. Is there any point I miss?

DDBM/ddbm/karras_diffusion.py

Lines 21 to 23 in 0d3cc96

	def vp_logs(t, beta_d, beta_min):
	t = th.as_tensor(t)
	return -0.25 * t ** 2 * (beta_d) - 0.5 * t * beta_min