pystra.active_learning.stopping.AllCriteria

Use it: User guide · Worked example · Theory

class AllCriteria(*, criteria)

Bases: StoppingCriterion

Require every supplied policy’s surrogate and sampling decisions.

Pass a nonempty tuple of stateless criteria. To reproduce the review’s combined beta rule, supply BetaBounds(consecutive=2) and BetaStability(consecutive=2). Each policy retains its sampling requirement; the final estimate must pass all of them.

Methods

accepts_estimate	Require all component sampling checks without recomputing uncertainty.
should_stop	Require all component policies on the same complete fit history.

property requires_bootstrap: bool

Whether any composed rule needs bootstrap probability ranges.

should_stop(history)

Require all component policies on the same complete fit history.

Return type:

bool

accepts_estimate(estimate)

Require all component sampling checks without recomputing uncertainty.

Return type:

bool