Interface RaptorTreeConfig

Configuration for the RAPTOR tree.

Interface

RaptorTreeConfig

interface RaptorTreeConfig {
    llmCaller: ((prompt) => Promise<string>);
    embeddingManager: IEmbeddingManager;
    vectorStore: IVectorStore;
    collectionName?: string;
    clusterSize?: number;
    maxDepth?: number;
    minChunksForLayer?: number;
}

Properties

llmCaller embeddingManager vectorStore collectionName? clusterSize? maxDepth? minChunksForLayer?

Properties

llmCaller

llmCaller: ((prompt) => Promise<string>)

LLM caller for generating summaries. Takes a prompt string and returns the LLM completion.

Type declaration

    • (prompt): Promise<string>

    • Parameters

      • prompt: string

        The full prompt including chain-of-thought instructions.

      Returns Promise<string>

Returns

The generated summary text.

embeddingManager

embeddingManager: IEmbeddingManager

Embedding manager for generating embeddings for clustering and storage.

vectorStore

vectorStore: IVectorStore

Vector store for persisting all layers of the tree.

Optional collectionName

collectionName?: string

Collection name in the vector store. Default: 'raptor-tree'.

Optional clusterSize

clusterSize?: number

Number of chunks per cluster. Default: 8.

Optional maxDepth

maxDepth?: number

Maximum tree depth (layers above leaf). Default: 4.

Optional minChunksForLayer

minChunksForLayer?: number

Minimum number of chunks required to form a new summary layer. Default: 3.

Settings

Member Visibility

Theme

On This Page

llmCallerembeddingManagervectorStorecollectionNameclusterSizemaxDepthminChunksForLayer