@maschinenlesbar.org/openka-cli
    Preparing search index...

    Interface EmbeddingSet

    Semantic search vectors. They are precomputed in the factory and frozen — the line never embeds anything at runtime, which is why the model that produced them is recorded here by name and hash rather than being callable.

    interface EmbeddingSet {
        dimensions: number;
        model: string;
        model_sha256?: string;
        vectors: Record<string, number[]>;
    }
    Index
    dimensions: number
    model: string
    model_sha256?: string
    vectors: Record<string, number[]>

    Document id -> unit-length vector.