gfa.tracking#

Fit tracking and serialization for GFA models.

Classes

`GFAFitTracker`	Quantities tracked during a GFA model fit.
`GFAFitFlags`	Status flags from a GFA model fit.
`GFAFitContext`	Context passed to callbacks during GFA fitting.

class GFAFitTracker( lb: ndarray | None = None, iter_time: ndarray | None = None, lb_base: float | None = None, )[source]#

Quantities tracked during a GFA model fit.

Attributes:

lbndarray of float, shape (num_iter,): Variational lower bound at each iteration.
iter_timendarray of float, shape (num_iter,): Runtime on each iteration.
lb_basefloat or None: Baseline lower bound for convergence checking.

class GFAFitFlags( converged: bool = False, decreasing_lb: bool = False, private_var_floor: bool = False, x_dims_removed: int = 0, )[source]#

Status flags from a GFA model fit.

Parameters:

convergedbool, default False: True if the lower bound converged before reaching max_iter.
decreasing_lbbool, default False: True if lower bound decreased during fitting.
private_var_floorbool, default False: True if the private variance floor was used on any values of phi.
x_dims_removedint, default 0: Number of latent dimensions removed due to low variance.

class GFAFitContext( config: GFAFitConfig, obs_hyperprior: ObsParamsHyperPrior, obs_posterior: ObsParamsPosterior, latents_posterior: LatentsPosteriorStatic, tracker: GFAFitTracker, flags: GFAFitFlags, )[source]#

Context passed to callbacks during GFA fitting.

Provides read-only access to fitting state and a save() method for checkpointing. Checkpoints can be loaded via GFAModel.load().

Parameters:

configGFAFitConfig: Fitting configuration.
obs_hyperpriorObsParamsHyperPrior: Prior hyperparameters.
obs_posteriorObsParamsPosterior: Observation model posterior.
latents_posteriorLatentsPosteriorStatic: Latent variable posterior.
trackerGFAFitTracker: Fitting progress tracker.
flagsGFAFitFlags: Fitting status flags.