cgpm
cgpm copied to clipboard
probcomp
Library of composable generative population models which serve as the modeling and inference backend of BayesDB.
`check.sh` passes, although only the `else` condition. Something in `pytest` changed, where filtering out the integration tests no longer works. (I've commented out related code in a couple of the...
This fixes install from git via pip on Python 3 for Coltrane.
We got a requirement by our industrial collaborator to bump the version of sqlalchemy. sqlalchemy is a dependency of CGPM. I tracked this down to CGPM through our build system...
Specifically the following lines https://github.com/probcomp/cgpm/blob/46e1d8d931fa81b2f4c3b4ff4575d9f8da1ef635/src/primitives/poisson.py#L146-L170 The posterior predictive distribution should be a negative binomial with n = a' and p = 1 / (1 + b'), or possibly p =...
In normal.py, the comment says `rho ~ Gamma(nu/2, s/2)` but the code says `rho = rng.gamma(nu/2., scale=2./s)` This is confusing. Either the comment is wrong, or else the presentation would...
This patch is needed since the column names I am getting from loom while calling `_update_state` (and `_retrieve_featureid_to_cgpm`, specifically) don't follow the assumptions made in the code. In fact, the...
TODO: Write design issues -- short version is that logpdf estimators are hard!
Therefore, every output SP will have the form ``` assume sim_output = mem((rowaddr, inputs) -> {...}; ``` as opposed to the current convention ``` assume sim_output = mem((rowaddr, apogee_km, perigee_km)...