NexusLabs.Foundry.Evaluation.Experiments

NexusLabs.Foundry.Evaluation

NexusLabs.Foundry.Evaluation.Experiments Namespace

Classes
ExperimentAttemptResult	Describes one operational execution attempt.
ExperimentBinaryStatisticalEvidence	Describes binary-success proportion evidence and its uncertainty.
ExperimentBinarySuccessPolicy<TCase,TOutput>	Applies a one-sided Wilson score decision to a required boolean item metric.
ExperimentCase<TCase>	Defines one logical experiment case and the number of statistically independent trials to run.
ExperimentCaseSourceResult<TCase>	Contains a finite ordered case collection and its provider-neutral source identity.
ExperimentConcurrencyLimiter	Provides a semaphore-backed concurrency limiter suitable for sharing through dependency injection.
ExperimentDefinition<TCase,TOutput>	Defines a provider-neutral experiment's finite source and item task.
ExperimentDeterministicPolicyEvidence	Describes deterministic threshold evidence from one named run evaluation.
ExperimentFailure	Provides stable machine-actionable failure data without retaining a raw exception or stack.
ExperimentItemCorrelation	Provides one namespaced provider identifier associated with an experiment item.
ExperimentItemEvaluationContext<TCase,TOutput>	Provides a successful task output and its stable identity to an item evaluator.
ExperimentItemFeatureCollection	Provides exact-type access to adapter-owned features for one experiment item scope.
ExperimentItemPublicationOperationResult	Describes item-scope publication work before the runner applies registered provider identity.
ExperimentItemPublicationResult	Describes one item scope's publication outcome independently from item quality.
ExperimentItemResult<TCase,TOutput>	Describes one case trial, including its complete attempt history and terminal output.
ExperimentItemScopeContext<TCase>	Provides stable identity and case data when entering one per-trial item scope.
ExperimentItemStatusCount	Associates an item status with its observed count.
ExperimentJsonArtifactWriter	Writes the schema-versioned deterministic Foundry experiment JSON envelope.
ExperimentMetricDiagnosticSnapshot	Provides an immutable normalized metric diagnostic.
ExperimentMetricInterpretationSnapshot	Provides an immutable normalized metric interpretation.
ExperimentMetricSnapshot	Provides an immutable Foundry-owned projection of one MEAI evaluation metric.
ExperimentPolicyContext<TCase,TOutput>	Provides immutable run measurements to an experiment policy.
ExperimentPolicyResult	Describes one isolated experiment policy outcome.
ExperimentPolicyVerdict	Provides a policy's decision and evidence before runner-owned result identity is applied.
ExperimentRetryContext	Provides one completed execution attempt to a retry policy.
ExperimentRetryDecision	Describes whether and when a failed execution attempt should be retried.
ExperimentRetryPolicy	Provides a bounded retry policy with explicit outcome selection and delay calculation.
ExperimentRunEvaluationContext<TCase,TOutput>	Provides every ordered item and stable run identity to a run evaluator.
ExperimentRunEvaluationResult	Describes one isolated run-evaluation outcome.
ExperimentRunEvaluationThresholdPolicy<TCase,TOutput>	Applies reusable deterministic thresholds to one named run evaluation.
ExperimentRunEvaluator<TCase,TOutput>	Adapts a named run-evaluation delegate to IExperimentRunEvaluator<TCase,TOutput>.
ExperimentRunner	Default bounded-worker implementation of IExperimentRunner.
ExperimentRunOptions	Configures one provider-neutral experiment run.
ExperimentRunOutcome<TCase,TOutput>	Provides one canonical quality result plus independent provider publication outcomes.
ExperimentRunResult<TCase,TOutput>	Provides the canonical ordered result for one provider-neutral experiment run.
ExperimentSinkPublicationOperationResult	Describes result-sink publication work before the runner applies registered sink identity.
ExperimentSinkResult	Describes one final result sink's publication outcome independently from run quality.
ExperimentSourceReference	Identifies the provider-neutral source from which experiment cases were materialized.
ExperimentTaskContext<TCase>	Provides stable identity and case data to one experiment task attempt.
LocalExperimentCaseSource<TCase>	Provides a finite in-process experiment case collection.

Interfaces
IExperimentCaseSource<TCase>	Materializes a finite ordered experiment case collection before execution begins.
IExperimentConcurrencyLimiter	Provides a caller-owned concurrency boundary that can be shared across experiment runs.
IExperimentItemScope<TCase,TOutput>	Provides one provider lifecycle spanning every attempt and item evaluation for a trial.
IExperimentItemScopeProvider<TCase,TOutput>	Creates one provider-neutral lifecycle scope for each statistical trial.
IExperimentResultSink<TCase,TOutput>	Publishes a completed canonical experiment result without changing its quality decision.
IExperimentRetryPolicy	Selects bounded execution retries for experiment items.
IExperimentRunEvaluator<TCase,TOutput>	Measures a complete ordered experiment run.
IExperimentRunner	Runs finite provider-neutral experiments with bounded task concurrency.
IExperimentRunPolicy<TCase,TOutput>	Produces a structured decision from complete experiment measurements.

Enums
ExperimentAttemptStatus	Describes the terminal state of one operational experiment attempt.
ExperimentConfidenceIntervalMethod	Identifies the confidence interval used by a statistical policy.
ExperimentFailureCode	Identifies a stable experiment failure category.
ExperimentFailureStage	Identifies the experiment stage that produced a structured failure.
ExperimentItemScopeFailureMode	Defines whether an item-scope setup failure may prevent a task attempt from starting.
ExperimentItemStatus	Describes the terminal quality-processing state of one experiment case trial.
ExperimentMetricDiagnosticSeverity	Identifies the stable severity of a normalized metric diagnostic.
ExperimentMetricKind	Identifies the normalized value kind of an experiment metric snapshot.
ExperimentMetricNonFiniteValue	Identifies a non-finite numeric metric value that cannot be represented as a JSON number.
ExperimentMetricRating	Provides a stable provider-owned projection of MEAI metric ratings.
ExperimentPolicyKind	Identifies the evidence model used by an experiment policy.
ExperimentPublicationOperationStatus	Describes whether one item scope or final sink completed its publication work.
ExperimentPublicationStatus	Describes aggregate publication health across item scopes and final result sinks.
ExperimentRetryableOutcome	Identifies execution outcomes that a retry policy may select.
ExperimentRunDecision	Identifies the aggregate decision for an experiment run.
ExperimentRunEvaluationStatus	Identifies the terminal status of one run evaluator.
ExperimentUnknownSampleTreatment	Configures how unknown binary samples affect a statistical policy.

Delegates
ExperimentItemEvaluator<TCase,TOutput>(ExperimentItemEvaluationContext<TCase,TOutput>, CancellationToken)	Evaluates one terminal successful experiment item exactly once.
ExperimentRetryDelayProvider(ExperimentRetryContext)	Computes an explicit retry delay from a completed execution attempt.
ExperimentRunEvaluationHandler<TCase,TOutput>(ExperimentRunEvaluationContext<TCase,TOutput>, CancellationToken)	Measures a complete ordered experiment run.
ExperimentTask<TCase,TOutput>(ExperimentTaskContext<TCase>, CancellationToken)	Executes one operational attempt for one experiment case trial.