Evalkeeper
    Preparing search index...

    Function requestHashFor

    • What makes two requests identical, so the second can reuse the first's answer: the suite, model, reasoning setting, instructions, input, and repeat. One function, so tests and estimates always agree.

      Parameters

      • r: {
            effort: string | null;
            input: unknown;
            maxOutputTokens?: number | null;
            model: string;
            prompt: string;
            repeat: number;
            suite: string;
            temperature?: number | null;
        }
        • effort: string | null
        • input: unknown
        • OptionalmaxOutputTokens?: number | null

          Left out of the hash when null, like temperature.

        • model: string
        • prompt: string
        • repeat: number
        • suite: string
        • Optionaltemperature?: number | null

          Left out of the hash when null, so requests without one reuse as before.

      Returns string