This repo contains the Tensorflow implementation for paper Implicit learning of convective organization explains precipitation stochasticity
The code to train the network is named AE_Pw_NN_precip.py which has org variables as input. To switch to baseline network simply remove the z_orig from the precip_model input.
Org_NN has few options including training the decoder or adding RI loss. To change the set-up, json file needs to be up-dated.