Is it possible to save results after every epoch? so that if there is some issue with power i can continue my training from the same point.

i am finetunning alexnet on FER2013

i want to set epoch size 300 but it takes alot of time in which there is a chance that my pc will turn off due to power issue.

so is it possible that i save results or "something" after every epoch so that i can continue my trainning from where it was stoped.

thanks in advance.

Answer 1

[yanqi liu]

sir,may be use ref https://www.mathworks.com/help/releases/R2021a/deeplearning/ug/resume-training-from-a-checkpoint-network.html

checkpointPath = pwd;
options = trainingOptions('sgdm', ...
    'InitialLearnRate',0.1, ...
    'MaxEpochs',20, ...
    'Verbose',false, ...
    'Plots','training-progress', ...
    'Shuffle','every-epoch', ...
    'CheckpointPath',checkpointPath);

[Mamoon ul Hassan Jan]

Thankyou Mr.yanqi liu.

my issue is resolved .

acctuly i found an easy way of doing it , if you are using deep network designer tool box all you have to do is to give a path for checkpoint and check the check box of checkpoints in training options .

💕