Replies: 1 comment
-
|
您好,如果验证集比较简单的话确实可能出现过拟合的问题,可以降低下训练轮次,或者看下验证集中是否有些图片在训练集也出现了 |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
train数据集17000张,val验证集1880张。训练500轮,300轮左右训练acc已经是1.0,但是val为97%,后续val持续上升,400轮val98.0%,440轮98.2%,480轮98.5%,即最佳模型为480轮的,想请教下,按理来说这已经算是过拟合了吧,为什么val仍会持续上升,有办法优化这种过拟合现象吗,在配置文件中是开起了reccaug数据增强的。配置文件仅修改了字符表
Beta Was this translation helpful? Give feedback.
All reactions