insilicomedicine / GENTRL

Generative Tensorial Reinforcement Learning (GENTRL) model
611 stars 218 forks source link

KLDiv equation #30

Open aced125 opened 3 years ago

aced125 commented 3 years ago

https://github.com/insilicomedicine/GENTRL/blob/master/gentrl/gentrl.py#L85

        kldiv_part = (-normal_distr_hentropies - log_p_zy).mean()

Should the second term not be log_p_z_by_y?