Fullrank VI implementation #470
Replies: 3 comments 4 replies
-
Indeed, look at what we do in |
Beta Was this translation helpful? Give feedback.
-
@rlouf |
Beta Was this translation helpful? Give feedback.
-
@rlouf @junpenglao class FullrankVIState(NamedTuple):
mu: PyTree
L: Array What do you think? |
Beta Was this translation helpful? Give feedback.
-
In mean-field VI, the variational posterior for all latent variables are independent therefore we can directly utilize tree_map to compute the density https://github.com/blackjax-devs/blackjax/blob/main/blackjax/vi/meanfield_vi.py#L127
However, for fullrank VI, the variational posterior is a multivariate Gaussian over all latent variables, so I assume we have to first flatten all the latent variables in order to compute log_q, is that correct?
Beta Was this translation helpful? Give feedback.
All reactions