@mlc-ai/web-llm/lib/openai_api_protocols/embedding.d.ts

Hardware accelerated language model chats on browsers

/**
 * The input to OpenAI API, directly adopted from openai-node with small tweaks:
 * https://github.com/openai/openai-node/blob/master/src/resources/embeddings.ts
 *
 * Copyright 2024 OpenAI
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *      http://www.apache.org/licenses/LICENSE-2.0
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
import { MLCEngineInterface } from "../types";
export declare class Embeddings {
    private engine;
    constructor(engine: MLCEngineInterface);
    /**
     * Creates an embedding vector representing the input text.
     */
    create(request: EmbeddingCreateParams): Promise<CreateEmbeddingResponse>;
}
export interface CreateEmbeddingResponse {
    /**
     * The list of embeddings generated by the model.
     */
    data: Array<Embedding>;
    /**
     * The name of the model used to generate the embedding.
     */
    model: string;
    /**
     * The object type, which is always "list".
     */
    object: "list";
    /**
     * The usage information for the request.
     */
    usage: CreateEmbeddingResponse.Usage;
}
export declare namespace CreateEmbeddingResponse {
    /**
     * The usage information for the request.
     */
    interface Usage {
        /**
         * The number of tokens used by the prompt.
         */
        prompt_tokens: number;
        /**
         * The total number of tokens used by the request.
         */
        total_tokens: number;
        /**
         * Fields specific to WebLLM, not present in OpenAI.
         */
        extra: {
            /**
             * Number of tokens per second for prefilling.
             */
            prefill_tokens_per_s: number;
        };
    }
}
/**
 * Represents an embedding vector returned by embedding endpoint.
 */
export interface Embedding {
    /**
     * The embedding vector, which is a list of floats. The length of vector depends on
     * the model.
     */
    embedding: Array<number>;
    /**
     * The index of the embedding in the list of embeddings.
     */
    index: number;
    /**
     * The object type, which is always "embedding".
     */
    object: "embedding";
}
export interface EmbeddingCreateParams {
    /**
     * Input text to embed, encoded as a string or array of tokens. To embed multiple
     * inputs in a single request, pass an array of strings or array of token arrays.
     * The input must not exceed the max input tokens for the model, and cannot be an empty string.
     * If the batch size is too large, multiple forward of the will take place.
     */
    input: string | Array<string> | Array<number> | Array<Array<number>>;
    /**
     * ID of the model to use. This equals to `ModelRecord.model_id`, which needs to either be in
     * `webllm.prebuiltAppConfig` or in `engineConfig.appConfig`.
     *
     * @note Call `CreateMLCEngine(model)` or `engine.reload(model)` ahead of time.
     * @note If only one model is loaded in the engine, this field is optional. If multiple models
     *   are loaded, this is required.
     */
    model?: string | null;
    /**
     * The format to return the embeddings in.
     *
     * @note Currently only support `float`.
     */
    encoding_format?: "float" | "base64";
    /**
     * The number of dimensions the resulting output embeddings should have.
     *
     * @note Not supported.
     */
    dimensions?: number;
    /**
     * A unique identifier representing your end-user, which can help OpenAI to monitor
     * and detect abuse.
     *
     * @note Not supported.
     */
    user?: string;
}
export declare const EmbeddingCreateParamsUnsupportedFields: Array<string>;
export declare function postInitAndCheckFields(request: EmbeddingCreateParams, currentModelId: string): void;
//# sourceMappingURL=embedding.d.ts.map