ГЛОБАЛЬНЫЙ РЕФАКТОРИНГ АРХИТЕКТУРЫ - Stage 4.6: добавить метрику stage4_contract_compliance_rate в Stage1 eval и отчеты
This commit is contained in:
@@ -109,7 +109,8 @@ describe("stage1 contract scaffolding", () => {
|
||||
"false_confidence_rate",
|
||||
"broad_answer_rate",
|
||||
"mechanism_specificity_score",
|
||||
"followup_context_retention_score"
|
||||
"followup_context_retention_score",
|
||||
"stage4_contract_compliance_rate"
|
||||
]);
|
||||
for (const metric of metricNames) {
|
||||
const bands = ACCOUNTANT_SCORING_RUBRIC_V01[metric as keyof typeof ACCOUNTANT_SCORING_RUBRIC_V01];
|
||||
|
||||
@@ -84,7 +84,8 @@ describe.sequential("assistant Stage 1 eval harness", () => {
|
||||
"false_confidence_rate",
|
||||
"broad_answer_rate",
|
||||
"mechanism_specificity_score",
|
||||
"followup_context_retention_score"
|
||||
"followup_context_retention_score",
|
||||
"stage4_contract_compliance_rate"
|
||||
]);
|
||||
expect(response.body.report?.rubric_bands?.generic_explanation_rate).toBeTruthy();
|
||||
expect(response.body.report?.feature_profile_snapshot).toBeTruthy();
|
||||
|
||||
Reference in New Issue
Block a user