When I use your refinedet to train our data ,I find the loss is very big why ?

yqyao / SSD_Pytorch

support different SSDs and different scale test, support refineDet.

MIT License

148 stars 51 forks source link

Open lingtengqiu opened 5 years ago

lingtengqiu commented 5 years ago

for example the train loss for total_loss as to 4.0 ? why this condition happen? do you divide batch_size ? please help me

yqyao commented 5 years ago

At the begin of training, it's ok. After several epochs, the loss will decrease. @lingtengqiu

lingtengqiu commented 5 years ago

I try , at the begin of training the loss is 16 ,at the end of epoch the loss is 4.0

rw1995 commented 5 years ago

我尝试，在训练开始时损失是16，在时代结束时损失是4.0

请问一下，数据是如何合并的，我尝试了许多方法，但是没有一个正确的，请问您是如何详细操作的？ @lingtengqiu