f90 / FactorGAN

FactorGAN - Training GANs with missing data
MIT License
35 stars 3 forks source link

From #1: Checkpoint creation #3

Closed f90 closed 4 years ago

f90 commented 4 years ago

From issue #1 : @kalai2033 asks: I don't see any checkpoints created. There have been 3 epochs completed so far.

f90 commented 4 years ago

So far, the code is setup so that the model is only saved at the very end of training. Going to add this into code now.

f90 commented 4 years ago

Code is updated, checkpoints are now saved after each epoch, and you can evaluate a different model than the final one, e.g. "G_1" for the model after second epoch, by providing --eval --eval_model G_1 as commandline parameters. Be sure to set --experiment_name parameter properly though