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

损失函数 #19

Open
zhangtiantian12311 opened this issue May 18, 2020 · 16 comments
Open

损失函数 #19

zhangtiantian12311 opened this issue May 18, 2020 · 16 comments

Comments

@zhangtiantian12311
Copy link

请问总损失一直维持在2.2,是什么问题呢,参数改了调小了初始学习率,总损失应该维持在多少大小?

@zhangcj13
Copy link

训练损失epoch: 140 |loss 0.883822 | hm_loss 0.629608 | lm_loss 1.895640 | wh_loss 0.045801 | off_loss 0.060070 | time 18.666667 |,就是lm_loss 一直下不去,最后检测关键点也不是很准

@XinyingZheng
Copy link

@zhangtiantian12311
Copy link
Author

zhangtiantian12311 commented May 21, 2020 via email

@XinyingZheng
Copy link

XinyingZheng commented May 21, 2020 via email

@zhangtiantian12311
Copy link
Author

zhangtiantian12311 commented May 21, 2020 via email

@XinyingZheng
Copy link

XinyingZheng commented May 21, 2020 via email

@XinyingZheng
Copy link

XinyingZheng commented May 22, 2020 via email

@XinyingZheng
Copy link

XinyingZheng commented Jun 24, 2020 via email

@lihua-hua
Copy link

@XinyingZheng 我的邮箱是[email protected],谢谢

@iytChen
Copy link

iytChen commented Jul 5, 2020

@XinyingZheng 你好,我从epoch 3开始就loss就变成nan了,后续的epoch也都是nan,请问遇到这样的问题一般是什么原因呢?

@iytChen
Copy link

iytChen commented Jul 5, 2020

@zhangcj13 你好,我从epoch 3开始就loss就变成nan了,后续的epoch也都是nan,请问遇到这样的问题一般是什么原因呢?

@lihua-hua
Copy link

@iytChen 可能是学习率的问题,你可以参与考一下 #24

@XinyingZheng
Copy link

XinyingZheng commented Jul 6, 2020 via email

@wuxiaolianggit
Copy link

各位大神,我利用demo跑出来的dets,有很多点的坐标是负数,这个是怎么回事呢?有大佬知道吗 @lihua-hua @XinyingZheng @zhangcj13 @zhangtiantian12311 @iytChen

@wuxiaolianggit
Copy link

各位大神,demo跑出来的dets的范围也是超出了图片像素坐标的范围,这个算法感觉预测出来是不能用啊

@RRRRRayyyyy
Copy link

训练损失epoch: 140 |loss 0.883822 | hm_loss 0.629608 | lm_loss 1.895640 | wh_loss 0.045801 | off_loss 0.060070 | time 18.666667 |,就是lm_loss 一直下不去,最后检测关键点也不是很准

请问您这个lr 、bs、input_size 分别设置是多少呢?是继承的作者的预训练模型继续训练的吗?谢谢。

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

7 participants