2
$\begingroup$

another dumb question, but how do you save the progress an ML model has made and start from that point later? Its kind of a vague question, but this is an example of what I am talking about:

Say, hypothetically speaking, if I just trained a really, really good machine learning model, like 99% testing and training accuracy. The problem is it took me 8 hours to get to that point and I would like the model to START from that point without having to retrain it when I am using it for the future, how would I save that progress?

I am currently messing around with TensorFlow and would like to know how to save progress for a NN I've been training with it.

$\endgroup$
  • 1
    $\begingroup$ depends on the software $\endgroup$ – Sycorax says Reinstate Monica Feb 13 at 2:27
  • 2
    $\begingroup$ And on the procedure: the answer for a random forest would be different from that for a neural network. (I think asking how to store progress so far when using a particular approach would be on topic here.) $\endgroup$ – Scortchi - Reinstate Monica Feb 13 at 8:07
0
$\begingroup$

checkpoints are the solution, every deep learning frameworks has this option, keras, tensorflow or pytorch.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.