Expand description
Offline next-app training/evaluation helpers for LSApp-shaped datasets.
Structs§
- Eval
Options - Measured
NetBenefit Inputs - Inputs required to compute measured net benefit of a ranker.
- Measured
Value - A measured scalar input loaded from an evaluation fixture.
- Measurement
Source - Provenance for a measured value. A value may be conservative or derived from measured counters, but it must not be an unlabelled placeholder.
- NetBenefit
Inputs - Inputs required to compute the net benefit of a ranker.
- NetBenefit
Report - Result of a net-benefit computation.
- NetBenefit
Trace - Prewarm
Fixture Build Inputs - Inputs used to build a PreWarmProcess fixture from already-collected offline measurements.
- Prewarm
NetBenefit Fixture - Fully measured PreWarmProcess net-benefit fixture.
- Train
Options
Enums§
Functions§
- build_
prewarm_ net_ benefit_ fixture - compute_
measured_ net_ benefit - Compute net benefit from fixture-backed measured inputs.
- compute_
net_ benefit - Compute net benefit for
examplespredictions. - evaluate
- train