You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi @Parskatt, thank you for noticing this! I believe you are correct, and this was not intended. Practically, I believe this does not make a difference, but we need to rename the parameter.
Yeah, probably it doesn't matter since you initialize inv_std so that the softplus puts it at 1. Maybe its slightly easier to get a singular distribution (i.e. close to zero variance) with the covariance parameterization, don't think it should be too bad though :)
Right :) We are also typically not training these covariance parameters and just keep them fixed at 1. throughout. Definitely a mistake on our side though.
At:
flowgmm/flow_ssl/distributions.py
Lines 22 to 25 in 422ff5d
You seem to insert a precision matrix, but reading the documentation:
https://github.com/pytorch/pytorch/blob/8b248af35d43c97d0e437f6f4ff0fbd4da5700c8/torch/distributions/multivariate_normal.py#L119
It seems like if you do not specify the matrix as a precision matrix, it will be assumed to be a covariance matrix (by argument order).
Is this intended?
The text was updated successfully, but these errors were encountered: