MATLAB Answers

How can I extract metrics/data of the LSTM training progress ?

38 views (last 30 days)
Pierre Guy
Pierre Guy on 10 Oct 2019
Commented: Pierre Guy on 16 Oct 2019
While monitoring a deep learning (LSTM) training progress, how can I extract the quantities data (Training root mean square error (RMSE) , Smoothed training accuracy, Validation accuracy, Training loss, smoothed training loss, and validation loss ) at each iteration ?( suppose you want to export these data to excel).
Indeed, while training a deep (LSTM) network in matab, '' when you specify 'training-progress' as the 'Plots' value in trainingOptions and start network training, trainNetwork creates a figure and displays training metrics at every iteration. Each iteration is an estimation of the gradient and an update of the network parameters. If you specify validation data in trainingOptions, then the figure shows validation metrics each time trainNetwork validates the network. The figure plots the following: Training root mean square error (RMSE) , Smoothed training accuracy, Validation accuracy, Training loss, smoothed training loss, and validation loss."
Thank you.
Best regards,

  0 Comments

Sign in to comment.

Accepted Answer

Jalaj Gambhir
Jalaj Gambhir on 14 Oct 2019
Hi,
You can use
[net,info] = trainNetwork(XTrain,YTrain,layers,options);
while training a network, to access 'info' struct. This struct contains information such as 'TrainingLoss', 'TrainingRMSE' and 'BaseLearnRate' for each iteration of training. You can refer to the documentation for more information.

  1 Comment

Sign in to comment.

More Answers (0)

Sign in to answer this question.