Skip to main content

createEvalHarness

@forge/monorepo


@forge/monorepo / backend/src / createEvalHarness

Function: createEvalHarness()

createEvalHarness(deps): object

Defined in: backend/src/evaluation/index.ts:65

Parameters

deps

HarnessDeps

Returns

score()

score(context, input): Promise<HarnessRunResult>

Score every case and store the run.

The whole dataset in one run, because a partial run cannot gate a release and a run that skipped what it could not grade would gate on a shrinking subset without saying so.

Parameters

context

ExecutionContext

input
cases

readonly EvalCase[]

release

string

Returns

Promise<HarnessRunResult>