vmf_vae_nlp
vmf_vae_nlp copied to clipboard
Sampling from prior with kappa = 0
Hi, I am not sure how you sample from the prior at test time. For the sampling from the prior, which is a vMF with kappa = 0
, do you use vmf_batch.py
? If I understood correctly it should not be dependent on mu
while kappa = 0
. According to the formulation of z = w*mu+v*(1-w^2)^(1/2)
, w
must be zero, while I think it is not necessarily equal to zero in vmf_batch.py
. Do you use another function?