@axflow/models/dist/cohere/embedding.d.mts

Zero-dependency, modular SDK for building robust natural language applications

type SharedRequestOptions = {
    apiKey?: string;
    apiUrl?: string;
    fetch?: typeof fetch;
    headers?: Record<string, string>;
    signal?: AbortSignal;
};

declare namespace CohereEmbeddingTypes {
    type Request = {
        texts: string[];
        model?: string;
        truncate?: 'NONE' | 'START' | 'END';
        input_type?: string;
    };
    type RequestOptions = SharedRequestOptions;
    type Response = {
        id: string;
        embeddings: number[][];
        texts: string[];
        meta: {
            api_version: {
                version: string;
                is_deprecated?: boolean;
                is_experimental?: boolean;
            };
            warnings: string[];
        };
    };
}
/**
 * Calculate text embeddings using the Cohere API.
 *
 * @see https://docs.cohere.com/reference/embed
 *
 * @param request The request body sent to Cohere. See Cohere's documentation for /v1/embed for supported parameters.
 * @param options
 * @param options.apiKey Cohere API key.
 * @param options.apiUrl The url of the Cohere (or compatible) API. Defaults to https://api.cohere.ai/v1/embed.
 * @param options.fetch A custom implementation of fetch. Defaults to globalThis.fetch.
 * @param options.headers Optionally add additional HTTP headers to the request.
 * @param options.signal An AbortSignal that can be used to abort the fetch request.
 * @returns An object consisting of the text embeddings and other metadata. See Cohere's documentation for /v1/embed.
 */
declare function run(request: CohereEmbeddingTypes.Request, options: CohereEmbeddingTypes.RequestOptions): Promise<CohereEmbeddingTypes.Response>;
/**
 * An object that encapsulates methods for calling the Cohere Embed API.
 */
declare class CohereEmbedding {
    static run: typeof run;
}

export { CohereEmbedding, CohereEmbeddingTypes };