search

nrontsis / PILCO

Bayesian Reinforcement Learning in Tensorflow
MIT License
313 stars 84 forks source link

Utils pr #17

Open kyr-pol opened 5 years ago

kyr-pol commented 5 years ago

This is a small pr, to add a save and a load utility function for pilco. This will allow a model to be stored, and be loaded later, to resume training, test further etc. I also moved all wrappers to a new file, utils.py, along with the save and load function, so that they can be imported whenever needed. The new functions are tested, so I consider it safe to merge, even though it might be possible to have the same functionality in a more elegant way (if for example we can get gpflow.saver to work as in here)

nrontsis commented 5 years ago

I'll try to change this so as to use gplfow's saver

HareshKarnan commented 5 years ago

This will be useful. thanks !

archielee commented 4 years ago

Hi @nrontsis, is the code from this PR still the best way to save PILCO models? Thanks!