Function createRefusalEvaluator

createRefusalEvaluator<
RecordType extends Record<string, unknown> = RefusalEvaluationRecord,
>(
args: RefusalEvaluatorArgs<RecordType>,
): ClassificationEvaluator<RecordType>
Creates a refusal evaluator function.

This function returns an evaluator that detects when an LLM refuses, declines, or avoids answering a user query. It is use-case agnostic: it only detects whether a refusal occurred, not whether the refusal was appropriate.
Type Parameters
- RecordType extends Record<string, unknown> = RefusalEvaluationRecord
Parameters
- args: RefusalEvaluatorArgs<RecordType>
  The arguments for creating the refusal evaluator.
  - Optionalchoices?: ClassificationChoicesMap
  - OptionalinputMapping?: ObjectMapping<RecordType>
    The mapping of the input to evaluate to the shape that the evaluator expects
  - model: LanguageModel
  - Optionalname?: string
  - OptionaloptimizationDirection?: OptimizationDirection
  - OptionalpromptTemplate?: PromptTemplate
  - Optionaltelemetry?: TelemetryConfig
Returns ClassificationEvaluator<RecordType>
An evaluator function that takes a RefusalEvaluationRecord and returns a classification result indicating whether the output is a refusal or an answer.
Example
```
const evaluator = createRefusalEvaluator({ model: openai("gpt-4o-mini") });
const result = await evaluator.evaluate({
  input: "What is the capital of France?",
  output: "I'm sorry, I can only help with technical questions.",
});
console.log(result.label); // "refused" or "answered"
```
- Defined in llm/createRefusalEvaluator.ts:53