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

复现相关问题 #51

Open
syGe888 opened this issue Mar 27, 2024 · 2 comments
Open

复现相关问题 #51

syGe888 opened this issue Mar 27, 2024 · 2 comments

Comments

@syGe888
Copy link

syGe888 commented Mar 27, 2024

作者您好,向您请教一个问题,复现实验结果时,发现框架很容易崩溃。如setting1的x4超分任务,如果按照论文中的4e-4学习率,发现训练最终崩溃;我尝试修改学习率为1e-4,发现最终能够收敛,但是在set5 Gaussian8数据集上的psnr只有31.97;再尝试调整学习率为2e-4,发现最终结果又崩溃了;想请问您一下,有什么训练策略可以解决框架崩溃的问题,达到论文的psnr效果吗?

@Algolzw
Copy link
Collaborator

Algolzw commented Mar 27, 2024

你好,这个版本框架确实容易崩溃,我当时是不断的从断点重训(即设置resume_state)直到结束。也许可以写一个脚本来判断当崩溃时自动重启训练。

@syGe888
Copy link
Author

syGe888 commented Mar 27, 2024

好嘞,万分感谢

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