How to save loss, rmse, mae, and mape in every training epoch?

7 次查看（过去 30 天）
a a
a a 2020-10-29
回答： Pratik 2024-12-12
Is there any suggestion on how to save the loss, rmse, mae, and mape in every training epoch? I want to compare them in condition of different parameters later.
Cheers
FYI I calculate the rmse, mae, and mape in the end like this:
net = trainNetwork(XTrain,YTrain,layers,options);
net = predictAndUpdateState(net,XTrain);
[net,YPred] = predictAndUpdateState(net,XTest);
YPred = sig(1)*YPred + mu(1);
YTest = dataTest(1,:);
rmse = sqrt(mean((YPred-YTest).^2))
mae = mean(abs(YPred-YTest))
mape = mean(abs((YPred-YTest)./YTest))*100

请先登录，再回答此问题。

回答（1 个）

Pratik
Pratik 2024-12-12
Hi,
To monitor the metrics such as loss, rmse and etc, training options can be used. Also built in metric object can be used to store the values to use later.
Please refer to the following documentation for more information:
www.mathworks.com/help/deeplearning/ref/trainingoptions.html#bu59f0q_sep_mw_7396cbb9-29fe-473a-87d7-6db858aefa0d

类别

在 Help Center 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by