Closed icefengzhe closed 3 years ago
用的什么数据集,哪个网络呢?
我也遇到这个问题,用的Bisenet和自己的数据集,dataloader是仿照cityscapes.py写的。 大佬要是方便的话能否加个微信指导一下,微信号:dukewilliam2010
已加你好友,莫浮
发自我的iPhone
------------------ Original ------------------ From: lovetian0910 <notifications@github.com> Date: Wed,Nov 11,2020 2:52 PM To: Tramac/awesome-semantic-segmentation-pytorch <awesome-semantic-segmentation-pytorch@noreply.github.com> Cc: noob <862628057@qq.com>, Author <author@noreply.github.com> Subject: Re: [Tramac/awesome-semantic-segmentation-pytorch] question about loss (#149)
我也遇到这个问题,用的Bisenet和自己的数据集,dataloader是仿照cityscapes.py写的。 大佬要是方便的话能否加个微信指导一下,微信号:dukewilliam2010
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
loss震荡比较厉害并且下降一点后就不再收敛,我调整了一些超参依旧没有解决问题,请问关于超参的调试你有什么好的建议?