Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

损失函数的问题 #68

Open
MrWan001 opened this issue Dec 11, 2023 · 2 comments
Open

损失函数的问题 #68

MrWan001 opened this issue Dec 11, 2023 · 2 comments

Comments

@MrWan001
Copy link

作者您好!!
文中提出的损失函数,是将实际预测的Xt-1和最大似然估计推导出的理想最优Xt-1做损失。

我在想可不可以用ground truth噪声采样得到的Xt-1来做损失,即用generate_random_states函数中的实际noises算出的score,传入reverse_sde_step函数中得到的Xt-1,来和网络预测的噪声采样得到的Xt-1做损失,这样合理吗?

用generate_random_states函数中的noises算出的score,传入reverse_sde_step函数中得到的Xt-1,和文中根据最大似然估计推导出的最优Xt-1有什么区别,这个问题想请教下您的看法。

@Algolzw
Copy link
Owner

Algolzw commented Dec 13, 2023

你好!我大概想了一下,利用score得到的Xt_1是forward过程中的path,这可能不同于maximum likelihood得到的path。这个想法比较有意思不过我也没有做过实验,你可以试一下 ^_^.

@MrWan001
Copy link
Author

好的,谢谢!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants