metabeta

Pretrained checkpoints for metabeta, a PyTorch library for amortized in-context posterior estimation for Generalized Linear Mixed Models (GLMMs).
The source code is available at github.com/adkipnis/metabeta.
Likelihood-specific checkpoints are automatically downloaded and cached on first use.

Checkpoint specifications

File	link function	target type	fixed effects	random effects
metabeta-normal.pt	identity	continuous	1 - 12	1 - 4
metabeta-bernoulli.pt	logistic	binary	1 - 12	1 - 4
metabeta-poisson.pt	log	count	1 - 12	1 - 4