Primary benefit of this is that it allows us to 1) Sample block neural autoregressive flows fitted with maximum likelihood, and 2) evaluate the log probability of block neural autoregressive flows fitted with variational inference. This can potentially also be used in other flows in the future.
Primary benefit of this is that it allows us to 1) Sample block neural autoregressive flows fitted with maximum likelihood, and 2) evaluate the log probability of block neural autoregressive flows fitted with variational inference. This can potentially also be used in other flows in the future.