-
Notifications
You must be signed in to change notification settings - Fork 476
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
天池语义分割代码求教 #14
Comments
train.py里面·的·@autocast()要删掉,但我不清楚是不是这个原因哈,为NaN的话你可以尝试调节下学习率啥的,或者是标签有问题 |
感谢!
…------------------ 原始邮件 ------------------
发件人: "Qu Yuli"<notifications@github.com>;
发送时间: 2021年2月21日(星期天) 上午10:03
收件人: "DLLXW/data-science-competition"<data-science-competition@noreply.github.com>;
抄送: "景峰"<jinghehehe@foxmail.com>; "Author"<author@noreply.github.com>;
主题: Re: [DLLXW/data-science-competition] 天池语义分割代码求教 (#14)
您好,我尝试使用您的baseline,现想去掉混合精度训练,如图,但是损失变为nan,不知道具体原因,请大佬解惑。
train.py里面·的·@autocast()要删掉,但我不清楚是不是这个原因哈,为NaN的话你可以尝试调节下学习率啥的,或者是标签有问题
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
您好,我尝试使用您的baseline,现想去掉混合精度训练,如图,但是损失变为nan,不知道具体原因,请大佬解惑。
The text was updated successfully, but these errors were encountered: