Open imsrgadich opened 6 years ago
Is it because when the code exit with error, the checkpoint file is not saved properly? and then when it restores it can't find the params?
I need to delete the previously created files ./results/temp/
and ./results/temp/tensorboard/
and it runs fine!
(the reason might be obvious, but I'm unable to figure this out.)
@imsrgadich in case you or anyone else is still stuck on this, you need to use gpflow.defer_build() and explicitly name your model so that the new session uses the same variables names. See the "Creating a GPFlow model" section in the example in the GPFlow repo: https://github.com/GPflow/GPflow/blob/master/doc/source/notebooks/monitor-tensorboard.ipynb
I was trying to understand how the
hist_path
andrestore_path
works. When the run the notebook for the first time it runs, but second time it givescheckpoint error
. Can you please explain what is happening here. thank you.