mlx-bun library API
    Preparing search index...

    Interface GenerationDefaults

    Request defaults a graph declares for chat generation. The server applies them under explicit request and server settings; nothing outside the model decides them by name.

    interface GenerationDefaults {
        enableThinking?: boolean;
        noThinkTemperatureCap: number;
    }
    Index
    enableThinking? noThinkTemperatureCap
    enableThinking?: boolean

    enable_thinking when neither the request nor the server chose; absent leaves the template's own default.

    noThinkTemperatureCap: number

    Upper bound on the configured sampling temperature for replies with thinking off (model authors publish one think-mode temperature and recommend a cooler one for direct replies).

    Settings
    Member Visibility

    On This Page
    Properties