A single question within a multi-question ClassifierSchema.
Signature:
export declare interface ClassifierQuestion
Properties
| Property | Type | Description |
|---|---|---|
| id | string | Unique identifier key in the returned ClassifierResult record. |
| normalizePrior | boolean | Whether to normalize against the empty-context ("[None]") prior. |
| options | ClassifierOption[] | Candidate options (required for categorical and ordinal; optional for binary). |
| prompt | string | Task instruction or natural language condition prompt. |
| temperature | number | Softmax temperature override (> 0.0). When unset or <= 0, the model's calibrated temperature is used. |
| threshold | number | Decision threshold in [0.0, 1.0] for binary questions. Defaults to 0.5. |
| type | ClassifierQuestionType | Question evaluation type. |
ClassifierQuestion.id
Unique identifier key in the returned ClassifierResult record.
Signature:
id: string;
ClassifierQuestion.normalizePrior
Whether to normalize against the empty-context ("[None]") prior.
Signature:
normalizePrior?: boolean;
ClassifierQuestion.options
Candidate options (required for categorical and ordinal; optional for binary).
Signature:
options?: ClassifierOption[];
ClassifierQuestion.prompt
Task instruction or natural language condition prompt.
Signature:
prompt: string;
ClassifierQuestion.temperature
Softmax temperature override (> 0.0). When unset or <= 0, the model's calibrated temperature is used.
Signature:
temperature?: number;
ClassifierQuestion.threshold
Decision threshold in [0.0, 1.0] for binary questions. Defaults to 0.5.
Signature:
threshold?: number;
ClassifierQuestion.type
Question evaluation type.
Signature:
type: ClassifierQuestionType;