Skip to content

Graphorin API reference v0.7.0


Graphorin API reference / @graphorin/observability / / EvalReport

Interface: EvalReport<I, O>

Defined in: packages/observability/src/eval/types.ts:76

Final report shape.

Stable

Type Parameters

Type Parameter
I
O

Properties

PropertyModifierTypeDescriptionDefined in
aborted?readonlybooleantrue when the run was cut short by an aborted signal - results and summary then cover only the cases that finished before the abort (a partial report). Absent on a normal full run. See runEvals.packages/observability/src/eval/types.ts:110
resultsreadonlyreadonly EvalCaseResult<I, O>[]-packages/observability/src/eval/types.ts:77
summaryreadonly{ avgDurationMs: number; byScorer: Readonly<Record&lt;string, { avgScore: number | null; failed: number; passed: number; }&gt;>; failed: number; passed: number; passHatK?: { baseCases: number; k: number; value: number; }; passRateCi?: { hi: number; lo: number; }; total: number; }-packages/observability/src/eval/types.ts:78
summary.avgDurationMsreadonlynumber-packages/observability/src/eval/types.ts:82
summary.byScorerreadonlyReadonly<Record&lt;string, { avgScore: number | null; failed: number; passed: number; }&gt;>-packages/observability/src/eval/types.ts:83
summary.failedreadonlynumber-packages/observability/src/eval/types.ts:81
summary.passedreadonlynumber-packages/observability/src/eval/types.ts:80
summary.passHatK?readonly{ baseCases: number; k: number; value: number; }pass^k stability metric - fraction of base cases whose EVERY repeat iteration passed. Present only when the run used iterations > 1.packages/observability/src/eval/types.ts:99
summary.passHatK.baseCasesreadonlynumber-packages/observability/src/eval/types.ts:101
summary.passHatK.kreadonlynumber-packages/observability/src/eval/types.ts:100
summary.passHatK.valuereadonlynumber-packages/observability/src/eval/types.ts:102
summary.passRateCi?readonly{ hi: number; lo: number; }95% Wilson score interval on the overall pass rate (E8 / evals-05). Always present on reports produced by runEvals; optional so older persisted reports keep parsing.packages/observability/src/eval/types.ts:94
summary.passRateCi.hireadonlynumber-packages/observability/src/eval/types.ts:94
summary.passRateCi.loreadonlynumber-packages/observability/src/eval/types.ts:94
summary.totalreadonlynumber-packages/observability/src/eval/types.ts:79