为什么训练时不管用lstm还是transformer每个epoch的loss都完全一致呢？

KittenCN / stock_prediction

基于神经网络的通用股票预测模型 A general stock prediction model based on neural networks

https://www.coderfan.com

GNU General Public License v3.0

178 stars 50 forks source link

Open Darylmeng opened 7 months ago

Darylmeng commented 7 months ago

每个epoch的loss变化都是一样的，可视化出来就是每个epoch的loss曲线都在重复

KittenCN commented 7 months ago

我自己没有这个问题，lstm和transformer有明显的区别