mbi.callbacks

Defines callback mechanisms for monitoring optimization processes.

This module provides a Callback class that can be used to track and log various metrics during iterative algorithms, such as those used in estimating marginals. It logs loss values and other relevant statistics.

Functions

default	Creates a default Callback with standard loss functions.
log	Log a message using the configured log function.
set_log_fn	Override the library-wide log function (default: print).

Classes

Callback

mbi.callbacks.set_log_fn(fn)

Override the library-wide log function (default: print).

mbi.callbacks.log(*args, **kwargs)

Log a message using the configured log function.

class mbi.callbacks.Callback(loss_fns: dict[str, mbi.marginal_loss.MarginalLossFn], _call: int = 0, _logs: list = <factory>)

Bases: object

loss_fns: dict[str, MarginalLossFn]

property summary: dict[str, list]

mbi.callbacks.default(measurements: list[LinearMeasurement], domain: Domain, data: Projectable | None = None, include_primal_feasibility: bool | None = None) → Callback

Creates a default Callback with standard loss functions.