You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
随机裁剪64→256的patch对发自我的手机-------- 原始邮件 --------发件人: One1209 ***@***.***>日期: 2024年4月20日周六 晚上10:28收件人: XY-boy/EDiffSR ***@***.***>抄送: Subscribed ***@***.***>主 题: Re: [XY-boy/EDiffSR] setting.ymal文件中的设置? (Issue #14)
我在我的3090GPU上,setting.ymal文件中的设置为: GT_size: 256,LR_size: 64的话可以正常跑,但是设置为512和128的话就爆显存了,但是我的数据集和作者是一致的,也就是128到512的SR任务,不太明白如果setting.ymal文件中的设置为: GT_size: 256,LR_size: 64的话,是会对训练数据集再降采样为原来的一半吗?
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
为什么setting.ymal文件中的设置为: GT_size: 256,LR_size: 64
作者在论文中不是做的128到512的SR任务吗?
还有T和iterations的取舍,应该是一个高另一个就可以比较低吧?有什么参考的标准吗?
感谢您的杰出工作,希望收到您的回复!
The text was updated successfully, but these errors were encountered: