search

thangbui / geepee

A collection of Gaussian process models
MIT License
29 stars 11 forks source link

save/load models #2

Open thangbui opened 7 years ago

thangbui commented 7 years ago

Is saving/loading just the hyperparameters/parameters by calling get_hypers()/update_hypers() enough?

also create an example for doing so

g7dhaliwal commented 6 years ago

Hi,

I tried the save/load function. On loading my model is showing very high variance in the DGP predictions. I used aep.SDGPR function to train my model and saved it in pickle format.

On loading the model in a different file, I am using the following commands

model = aep.SDGPR(X, y_train_normalized, M,hidden_size, lik='Gaussian') model.load_model(fname = 'GeePeeModel/GBE-DGPModelHS1.pickle')

The above command results in a model with high GP variance.

Gurjot