Save Experiment without EarlyStopping

[Supersak80]

Hi all,
I would like to save an experiment (model, etc.) to disk after running it for the maximum number of epochs without the early stopping feature. How can I modify the convolutional neural network example to do this?

Thanks,
Sam

[eywalker]

At the end of your script (i.e. after xp:run(ds) completes), you can simply save the experiment object itself with

torch.save('/path/to/file', xp)

where xp is the experiment object. The early stopping feature is ultimately performing the same operation.