GithubHelp home page GithubHelp logo

Loss值 about rcf-pytorch HOT 10 OPEN

meteorshowers avatar meteorshowers commented on August 18, 2024
Loss值

from rcf-pytorch.

Comments (10)

lly2111101 avatar lly2111101 commented on August 18, 2024

您好,我用这个代码训练BSDS500数据loss非常的大6万多,模型训练的很慢且不收敛,想问一下您在训练时有没有遇到类似的问题,或者您觉得可能会是哪些原因导致的这种情况?

from rcf-pytorch.

mingweiY avatar mingweiY commented on August 18, 2024

from rcf-pytorch.

lly2111101 avatar lly2111101 commented on August 18, 2024

您好,我对这个也不太清楚,我没改代码,直接跑就是这个效果,您修改了什么吧?

------------------ 原始邮件 ------------------ 发件人: "lly2111101"<[email protected]>; 发送时间: 2019年9月30日(星期一) 下午4:13 收件人: "meteorshowers/RCF-pytorch"<[email protected]>; 抄送: "A Student"<[email protected]>; "Author"<[email protected]>; 主题: Re: [meteorshowers/RCF-pytorch] Loss值 (#31) 您好,我用这个代码训练BSDS500数据loss非常的大6万多,模型训练的很慢且不收敛,想问一下您在训练时有没有遇到类似的问题,或者您觉得可能会是哪些原因导致的这种情况? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

您好,我就是在数据的data_loader部分做了一点点的改动,只是为了读取数据,有可能是数据这部分出了问题?麻烦问一下您用的是BSDS500数据吗?可否分享一下您用的数据以及您使用的train.lst test.lst格式呢?

from rcf-pytorch.

rechealzrq avatar rechealzrq commented on August 18, 2024

您好,我对这个也不太清楚,我没改代码,直接跑就是这个效果,您修改了什么吧?

------------------原始邮件- -----------------发件人:“ lly2111101” < [email protected] >; 发送时间:2019年9月30日(星期一)下午4:13收件人:“ meteorshowers / RCF-pytorch” < [email protected] >; 抄送:“一个学生” < [email protected] >; “作者” < [email protected] >; 主题:Re:[meteorshowers / RCF-pytorch] Loss值(#31)您好,我用这个代码训练BSDS500数据损失非常的大6万多,模型训练的很慢且不收敛,想问一下您在训练-因为您编写了该线程,因此收到此消息是因为您当时收到了类似的问题,或者您发现可能会是某种原因导致的这种情况?直接回复此电子邮件,在GitHub上查看,或使该线程静音。

您好,我就是在数据的data_loader部分做到一点点的缩小,只是为了读取数据,有可能是数据这部分出了问题?麻烦问一下您用的是BSDS500数据吗?可否分享一下您用的数据以及您使用的train.lst test.lst格式​​呢?

请问下你的问题解决了吗?我也遇到了同样的问题

from rcf-pytorch.

AishuaiYao avatar AishuaiYao commented on August 18, 2024

@lly2111101 嘿嘿,同学您好,所以最后解决了吗,我现在也是这个问题,希望您能帮助我运行起来

from rcf-pytorch.

AishuaiYao avatar AishuaiYao commented on August 18, 2024

image
得到如图所示的loss,感觉这个loss降得好快啊,第一次跑训练不太懂是不是哪里搞错了,还是这个原本就是这样,感谢解答。

嗨,同学您好,我最近也在研究这个代码,但是出现了大家都遇到的问题,我看issue上也没有人具体说一下怎么处理,我想请您指点一下,你的处理过程可以吗,谢谢

from rcf-pytorch.

mingweiY avatar mingweiY commented on August 18, 2024

from rcf-pytorch.

AishuaiYao avatar AishuaiYao commented on August 18, 2024

您好,好像loss曲线就是这样的,中间这些比较大波动的地方是每一个epoch的开始,因为就算了一张图像的loss,所以波动大,这是正常的,loss的计算方式导致的。好像就是收敛的挺快的,我就这么训练的,最后效果还不错。

------------------ 原始邮件 ------------------ 发件人: "meteorshowers/RCF-pytorch" @.>; 发送时间: 2021年4月1日(星期四) 中午11:08 @.>; 抄送: "A @.@.>; 主题: Re: [meteorshowers/RCF-pytorch] Loss值 (#31) 得到如图所示的loss,感觉这个loss降得好快啊,第一次跑训练不太懂是不是哪里搞错了,还是这个原本就是这样,感谢解答。 嗨,同学您好,我最近也在研究这个代码,但是出现了大家都遇到的问题,我看issue上也没有人具体说一下怎么处理,我想请您指点一下,你的处理过程可以吗,谢谢 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

哈哈,您好,没想到您回复的这么及时,嗯,其实我是想知道,你的处理流程,因为我觉得可能是我的处理上和您有什么区别。其实一个工程无非三个部分,输入,计算,输出。所以,我想知道

您的输入数据是下载了哪些文件,然后又是怎么预处理的。
然后就是计算,您的代码就是直接用的他们的原始参数吗,就是arg. 那部分的参数设定,什么都没改。
最后输出,您的输出可以观察到预期的边缘分割效果吗?
嗯,如果您愿意的话,可以不可以加一下我的微信,id:iwanttokeepadog,可以更方便的请教您
十分感谢

from rcf-pytorch.

1dhuh avatar 1dhuh commented on August 18, 2024

image
你好请问大概需要训练多少给epoch,我更换了数据集导致loss降低异常缓慢

from rcf-pytorch.

DeepDarkFantasy0-1-1 avatar DeepDarkFantasy0-1-1 commented on August 18, 2024

image 得到如图所示的loss,感觉这个loss降得好快啊,第一次跑训练不太懂是不是哪里搞错了,还是这个原本就是这样,感谢解答。

您好,请问这个loss函数的统计是加在哪里的?是train的def train里吗

from rcf-pytorch.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.