modelfusion 0.37.0 → 0.39.0

package/model-function/index.js

@@ -7,9 +7,9 @@ export * from "./SuccessfulModelCall.js";
 export * from "./describe-image/ImageDescriptionEvent.js";
 export * from "./describe-image/ImageDescriptionModel.js";
 export * from "./describe-image/describeImage.js";
-export * from "./embed-text/TextEmbeddingEvent.js";
-export * from "./embed-text/TextEmbeddingModel.js";
-export * from "./embed-text/embedText.js";
+export * from "./embed/EmbeddingEvent.js";
+export * from "./embed/EmbeddingModel.js";
+export * from "./embed/embed.js";
 export * from "./generate-image/ImageGenerationEvent.js";
 export * from "./generate-image/ImageGenerationModel.js";
 export * from "./generate-image/generateImage.js";

package/model-provider/cohere/CohereTextEmbeddingModel.cjs

@@ -5,9 +5,9 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.CohereTextEmbeddingModel = exports.COHERE_TEXT_EMBEDDING_MODELS = void 0;
 const zod_1 = __importDefault(require("zod"));
-const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
 const callWithRetryAndThrottle_js_1 = require("../../core/api/callWithRetryAndThrottle.cjs");
 const postToApi_js_1 = require("../../core/api/postToApi.cjs");
+const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
 const CohereApiConfiguration_js_1 = require("./CohereApiConfiguration.cjs");
 const CohereError_js_1 = require("./CohereError.cjs");
 const CohereTokenizer_js_1 = require("./CohereTokenizer.cjs");
@@ -31,7 +31,7 @@ exports.COHERE_TEXT_EMBEDDING_MODELS = {
  * @see https://docs.cohere.com/reference/embed
  *
  * @example
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   new CohereTextEmbeddingModel({ model: "embed-english-light-v2.0" }),
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -48,7 +48,7 @@ class CohereTextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
             writable: true,
             value: "cohere"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -94,8 +94,8 @@ class CohereTextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
         return this.tokenizer.detokenize(tokens);
     }
     async callAPI(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The Cohere embedding API only supports ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The Cohere embedding API only supports ${this.maxValuesPerCall} texts per API call.`);
         }
         const run = options?.run;
         const settings = options?.settings;

package/model-provider/cohere/CohereTextEmbeddingModel.d.ts

@@ -1,8 +1,8 @@
 import z from "zod";
-import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
+import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { ModelFunctionOptions } from "../../model-function/ModelFunctionOptions.js";
-import { TextEmbeddingModel, TextEmbeddingModelSettings } from "../../model-function/embed-text/TextEmbeddingModel.js";
+import { EmbeddingModel, EmbeddingModelSettings } from "../../model-function/embed/EmbeddingModel.js";
 import { FullTokenizer } from "../../model-function/tokenize-text/Tokenizer.js";
 export declare const COHERE_TEXT_EMBEDDING_MODELS: {
     "embed-english-light-v2.0": {
@@ -19,7 +19,7 @@ export declare const COHERE_TEXT_EMBEDDING_MODELS: {
     };
 };
 export type CohereTextEmbeddingModelType = keyof typeof COHERE_TEXT_EMBEDDING_MODELS;
-export interface CohereTextEmbeddingModelSettings extends TextEmbeddingModelSettings {
+export interface CohereTextEmbeddingModelSettings extends EmbeddingModelSettings {
     api?: ApiConfiguration;
     model: CohereTextEmbeddingModelType;
     truncate?: "NONE" | "START" | "END";
@@ -30,7 +30,7 @@ export interface CohereTextEmbeddingModelSettings extends TextEmbeddingModelSett
  * @see https://docs.cohere.com/reference/embed
  *
  * @example
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   new CohereTextEmbeddingModel({ model: "embed-english-light-v2.0" }),
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -38,11 +38,11 @@ export interface CohereTextEmbeddingModelSettings extends TextEmbeddingModelSett
  *   ]
  * );
  */
-export declare class CohereTextEmbeddingModel extends AbstractModel<CohereTextEmbeddingModelSettings> implements TextEmbeddingModel<CohereTextEmbeddingResponse, CohereTextEmbeddingModelSettings>, FullTokenizer {
+export declare class CohereTextEmbeddingModel extends AbstractModel<CohereTextEmbeddingModelSettings> implements EmbeddingModel<string, CohereTextEmbeddingResponse, CohereTextEmbeddingModelSettings>, FullTokenizer {
     constructor(settings: CohereTextEmbeddingModelSettings);
     readonly provider: "cohere";
     get modelName(): "embed-english-light-v2.0" | "embed-english-v2.0" | "embed-multilingual-v2.0";
-    readonly maxTextsPerCall = 96;
+    readonly maxValuesPerCall = 96;
     readonly embeddingDimensions: number;
     readonly contextWindowSize: number;
     private readonly tokenizer;

package/model-provider/cohere/CohereTextEmbeddingModel.js

@@ -1,7 +1,7 @@
 import z from "zod";
-import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { callWithRetryAndThrottle } from "../../core/api/callWithRetryAndThrottle.js";
 import { createJsonResponseHandler, postJsonToApi, } from "../../core/api/postToApi.js";
+import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { CohereApiConfiguration } from "./CohereApiConfiguration.js";
 import { failedCohereCallResponseHandler } from "./CohereError.js";
 import { CohereTokenizer } from "./CohereTokenizer.js";
@@ -25,7 +25,7 @@ export const COHERE_TEXT_EMBEDDING_MODELS = {
  * @see https://docs.cohere.com/reference/embed
  *
  * @example
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   new CohereTextEmbeddingModel({ model: "embed-english-light-v2.0" }),
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -42,7 +42,7 @@ export class CohereTextEmbeddingModel extends AbstractModel {
             writable: true,
             value: "cohere"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -88,8 +88,8 @@ export class CohereTextEmbeddingModel extends AbstractModel {
         return this.tokenizer.detokenize(tokens);
     }
     async callAPI(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The Cohere embedding API only supports ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The Cohere embedding API only supports ${this.maxValuesPerCall} texts per API call.`);
         }
         const run = options?.run;
         const settings = options?.settings;

package/model-provider/huggingface/HuggingFaceTextEmbeddingModel.cjs

@@ -22,7 +22,7 @@ const HuggingFaceApiConfiguration_js_1 = require("./HuggingFaceApiConfiguration.
  *   retry: retryWithExponentialBackoff({ maxTries: 5 }),
  * });
  *
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   model,
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -39,7 +39,7 @@ class HuggingFaceTextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
             writable: true,
             value: "huggingface"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -70,15 +70,15 @@ class HuggingFaceTextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
             value: undefined
         });
         // There is no limit documented in the HuggingFace API. Use 1024 as a reasonable default.
-        this.maxTextsPerCall = settings.maxTextsPerCall ?? 1024;
+        this.maxValuesPerCall = settings.maxValuesPerCall ?? 1024;
         this.embeddingDimensions = settings.embeddingDimensions;
     }
     get modelName() {
         return this.settings.model;
     }
     async callAPI(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The HuggingFace feature extraction API is configured to only support ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The HuggingFace feature extraction API is configured to only support ${this.maxValuesPerCall} texts per API call.`);
         }
         const run = options?.run;
         const settings = options?.settings;

package/model-provider/huggingface/HuggingFaceTextEmbeddingModel.d.ts

@@ -2,11 +2,11 @@ import z from "zod";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
 import { ModelFunctionOptions } from "../../model-function/ModelFunctionOptions.js";
-import { TextEmbeddingModel, TextEmbeddingModelSettings } from "../../model-function/embed-text/TextEmbeddingModel.js";
-export interface HuggingFaceTextEmbeddingModelSettings extends TextEmbeddingModelSettings {
+import { EmbeddingModel, EmbeddingModelSettings } from "../../model-function/embed/EmbeddingModel.js";
+export interface HuggingFaceTextEmbeddingModelSettings extends EmbeddingModelSettings {
     api?: ApiConfiguration;
     model: string;
-    maxTextsPerCall?: number;
+    maxValuesPerCall?: number;
     embeddingDimensions?: number;
     options?: {
         useCache?: boolean;
@@ -25,7 +25,7 @@ export interface HuggingFaceTextEmbeddingModelSettings extends TextEmbeddingMode
  *   retry: retryWithExponentialBackoff({ maxTries: 5 }),
  * });
  *
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   model,
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -33,11 +33,11 @@ export interface HuggingFaceTextEmbeddingModelSettings extends TextEmbeddingMode
  *   ]
  * );
  */
-export declare class HuggingFaceTextEmbeddingModel extends AbstractModel<HuggingFaceTextEmbeddingModelSettings> implements TextEmbeddingModel<HuggingFaceTextEmbeddingResponse, HuggingFaceTextEmbeddingModelSettings> {
+export declare class HuggingFaceTextEmbeddingModel extends AbstractModel<HuggingFaceTextEmbeddingModelSettings> implements EmbeddingModel<string, HuggingFaceTextEmbeddingResponse, HuggingFaceTextEmbeddingModelSettings> {
     constructor(settings: HuggingFaceTextEmbeddingModelSettings);
     readonly provider = "huggingface";
     get modelName(): string;
-    readonly maxTextsPerCall: number;
+    readonly maxValuesPerCall: number;
     readonly contextWindowSize: undefined;
     readonly embeddingDimensions: number | undefined;
     readonly tokenizer: undefined;

package/model-provider/huggingface/HuggingFaceTextEmbeddingModel.js

@@ -16,7 +16,7 @@ import { HuggingFaceApiConfiguration } from "./HuggingFaceApiConfiguration.js";
  *   retry: retryWithExponentialBackoff({ maxTries: 5 }),
  * });
  *
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   model,
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -33,7 +33,7 @@ export class HuggingFaceTextEmbeddingModel extends AbstractModel {
             writable: true,
             value: "huggingface"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -64,15 +64,15 @@ export class HuggingFaceTextEmbeddingModel extends AbstractModel {
             value: undefined
         });
         // There is no limit documented in the HuggingFace API. Use 1024 as a reasonable default.
-        this.maxTextsPerCall = settings.maxTextsPerCall ?? 1024;
+        this.maxValuesPerCall = settings.maxValuesPerCall ?? 1024;
         this.embeddingDimensions = settings.embeddingDimensions;
     }
     get modelName() {
         return this.settings.model;
     }
     async callAPI(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The HuggingFace feature extraction API is configured to only support ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The HuggingFace feature extraction API is configured to only support ${this.maxValuesPerCall} texts per API call.`);
         }
         const run = options?.run;
         const settings = options?.settings;

package/model-provider/llamacpp/LlamaCppTextEmbeddingModel.cjs

@@ -5,9 +5,9 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.LlamaCppTextEmbeddingModel = void 0;
 const zod_1 = __importDefault(require("zod"));
-const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
 const callWithRetryAndThrottle_js_1 = require("../../core/api/callWithRetryAndThrottle.cjs");
 const postToApi_js_1 = require("../../core/api/postToApi.cjs");
+const AbstractModel_js_1 = require("../../model-function/AbstractModel.cjs");
 const LlamaCppApiConfiguration_js_1 = require("./LlamaCppApiConfiguration.cjs");
 const LlamaCppError_js_1 = require("./LlamaCppError.cjs");
 const LlamaCppTokenizer_js_1 = require("./LlamaCppTokenizer.cjs");
@@ -20,7 +20,7 @@ class LlamaCppTextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
             writable: true,
             value: "llamacpp"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -54,8 +54,8 @@ class LlamaCppTextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
         return this.tokenizer.tokenize(text);
     }
     async callAPI(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The Llama.cpp embedding API only supports ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The Llama.cpp embedding API only supports ${this.maxValuesPerCall} texts per API call.`);
         }
         const run = options?.run;
         const settings = options?.settings;

package/model-provider/llamacpp/LlamaCppTextEmbeddingModel.d.ts

@@ -1,17 +1,17 @@
 import z from "zod";
-import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
+import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { ModelFunctionOptions } from "../../model-function/ModelFunctionOptions.js";
-import { TextEmbeddingModel, TextEmbeddingModelSettings } from "../../model-function/embed-text/TextEmbeddingModel.js";
-export interface LlamaCppTextEmbeddingModelSettings extends TextEmbeddingModelSettings {
+import { EmbeddingModel, EmbeddingModelSettings } from "../../model-function/embed/EmbeddingModel.js";
+export interface LlamaCppTextEmbeddingModelSettings extends EmbeddingModelSettings {
     api?: ApiConfiguration;
     embeddingDimensions?: number;
 }
-export declare class LlamaCppTextEmbeddingModel extends AbstractModel<LlamaCppTextEmbeddingModelSettings> implements TextEmbeddingModel<LlamaCppTextEmbeddingResponse, LlamaCppTextEmbeddingModelSettings> {
+export declare class LlamaCppTextEmbeddingModel extends AbstractModel<LlamaCppTextEmbeddingModelSettings> implements EmbeddingModel<string, LlamaCppTextEmbeddingResponse, LlamaCppTextEmbeddingModelSettings> {
     constructor(settings?: LlamaCppTextEmbeddingModelSettings);
     readonly provider: "llamacpp";
     get modelName(): null;
-    readonly maxTextsPerCall = 1;
+    readonly maxValuesPerCall = 1;
     readonly contextWindowSize: undefined;
     readonly embeddingDimensions: number | undefined;
     private readonly tokenizer;

package/model-provider/llamacpp/LlamaCppTextEmbeddingModel.js

@@ -1,7 +1,7 @@
 import z from "zod";
-import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { callWithRetryAndThrottle } from "../../core/api/callWithRetryAndThrottle.js";
 import { createJsonResponseHandler, postJsonToApi, } from "../../core/api/postToApi.js";
+import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { LlamaCppApiConfiguration } from "./LlamaCppApiConfiguration.js";
 import { failedLlamaCppCallResponseHandler } from "./LlamaCppError.js";
 import { LlamaCppTokenizer } from "./LlamaCppTokenizer.js";
@@ -14,7 +14,7 @@ export class LlamaCppTextEmbeddingModel extends AbstractModel {
             writable: true,
             value: "llamacpp"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -48,8 +48,8 @@ export class LlamaCppTextEmbeddingModel extends AbstractModel {
         return this.tokenizer.tokenize(text);
     }
     async callAPI(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The Llama.cpp embedding API only supports ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The Llama.cpp embedding API only supports ${this.maxValuesPerCall} texts per API call.`);
         }
         const run = options?.run;
         const settings = options?.settings;

package/model-provider/openai/OpenAICostCalculator.cjs

@@ -24,7 +24,7 @@ class OpenAICostCalculator {
                     settings: call.settings,
                 });
             }
-            case "text-embedding": {
+            case "embedding": {
                 if (model == null) {
                     return null;
                 }

package/model-provider/openai/OpenAICostCalculator.js

@@ -21,7 +21,7 @@ export class OpenAICostCalculator {
                     settings: call.settings,
                 });
             }
-            case "text-embedding": {
+            case "embedding": {
                 if (model == null) {
                     return null;
                 }

package/model-provider/openai/OpenAITextEmbeddingModel.cjs

@@ -37,7 +37,7 @@ exports.calculateOpenAIEmbeddingCostInMillicents = calculateOpenAIEmbeddingCostI
  * @see https://platform.openai.com/docs/api-reference/embeddings
  *
  * @example
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   new OpenAITextEmbeddingModel({ model: "text-embedding-ada-002" }),
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -54,7 +54,7 @@ class OpenAITextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
             writable: true,
             value: "openai"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -115,8 +115,8 @@ class OpenAITextEmbeddingModel extends AbstractModel_js_1.AbstractModel {
         return {};
     }
     generateEmbeddingResponse(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The OpenAI embedding API only supports ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The OpenAI embedding API only supports ${this.maxValuesPerCall} texts per API call.`);
         }
         return this.callAPI(texts, options);
     }

package/model-provider/openai/OpenAITextEmbeddingModel.d.ts

@@ -2,7 +2,7 @@ import z from "zod";
 import { AbstractModel } from "../../model-function/AbstractModel.js";
 import { ModelFunctionOptions } from "../../model-function/ModelFunctionOptions.js";
 import { ApiConfiguration } from "../../core/api/ApiConfiguration.js";
-import { TextEmbeddingModel, TextEmbeddingModelSettings } from "../../model-function/embed-text/TextEmbeddingModel.js";
+import { EmbeddingModel, EmbeddingModelSettings } from "../../model-function/embed/EmbeddingModel.js";
 import { TikTokenTokenizer } from "./TikTokenTokenizer.js";
 export declare const OPENAI_TEXT_EMBEDDING_MODELS: {
     "text-embedding-ada-002": {
@@ -17,7 +17,7 @@ export declare const calculateOpenAIEmbeddingCostInMillicents: ({ model, respons
     model: OpenAITextEmbeddingModelType;
     responses: OpenAITextEmbeddingResponse[];
 }) => number;
-export interface OpenAITextEmbeddingModelSettings extends TextEmbeddingModelSettings {
+export interface OpenAITextEmbeddingModelSettings extends EmbeddingModelSettings {
     api?: ApiConfiguration;
     model: OpenAITextEmbeddingModelType;
     isUserIdForwardingEnabled?: boolean;
@@ -28,7 +28,7 @@ export interface OpenAITextEmbeddingModelSettings extends TextEmbeddingModelSett
  * @see https://platform.openai.com/docs/api-reference/embeddings
  *
  * @example
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   new OpenAITextEmbeddingModel({ model: "text-embedding-ada-002" }),
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -36,11 +36,11 @@ export interface OpenAITextEmbeddingModelSettings extends TextEmbeddingModelSett
  *   ]
  * );
  */
-export declare class OpenAITextEmbeddingModel extends AbstractModel<OpenAITextEmbeddingModelSettings> implements TextEmbeddingModel<OpenAITextEmbeddingResponse, OpenAITextEmbeddingModelSettings> {
+export declare class OpenAITextEmbeddingModel extends AbstractModel<OpenAITextEmbeddingModelSettings> implements EmbeddingModel<string, OpenAITextEmbeddingResponse, OpenAITextEmbeddingModelSettings> {
     constructor(settings: OpenAITextEmbeddingModelSettings);
     readonly provider: "openai";
     get modelName(): "text-embedding-ada-002";
-    readonly maxTextsPerCall = 2048;
+    readonly maxValuesPerCall = 2048;
     readonly embeddingDimensions: number;
     readonly tokenizer: TikTokenTokenizer;
     readonly contextWindowSize: number;

package/model-provider/openai/OpenAITextEmbeddingModel.js

@@ -29,7 +29,7 @@ export const calculateOpenAIEmbeddingCostInMillicents = ({ model, responses, })
  * @see https://platform.openai.com/docs/api-reference/embeddings
  *
  * @example
- * const embeddings = await embedTexts(
+ * const embeddings = await embedMany(
  *   new OpenAITextEmbeddingModel({ model: "text-embedding-ada-002" }),
  *   [
  *     "At first, Nox didn't know what to do with the pup.",
@@ -46,7 +46,7 @@ export class OpenAITextEmbeddingModel extends AbstractModel {
             writable: true,
             value: "openai"
         });
-        Object.defineProperty(this, "maxTextsPerCall", {
+        Object.defineProperty(this, "maxValuesPerCall", {
             enumerable: true,
             configurable: true,
             writable: true,
@@ -107,8 +107,8 @@ export class OpenAITextEmbeddingModel extends AbstractModel {
         return {};
     }
     generateEmbeddingResponse(texts, options) {
-        if (texts.length > this.maxTextsPerCall) {
-            throw new Error(`The OpenAI embedding API only supports ${this.maxTextsPerCall} texts per API call.`);
+        if (texts.length > this.maxValuesPerCall) {
+            throw new Error(`The OpenAI embedding API only supports ${this.maxValuesPerCall} texts per API call.`);
         }
         return this.callAPI(texts, options);
     }

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "modelfusion",
   "description": "Build multimodal applications, chatbots, and agents with JavaScript and TypeScript.",
-  "version": "0.37.0",
+  "version": "0.39.0",
   "author": "Lars Grammel",
   "license": "MIT",
   "keywords": [

package/tool/Tool.d.ts

@@ -1,4 +1,5 @@
 import { FunctionOptions } from "../core/FunctionOptions.js";
+import { JsonSchemaProducer } from "../core/structure/JsonSchemaProducer.js";
 import { Schema } from "../core/structure/Schema.js";
 import { StructureDefinition } from "../core/structure/StructureDefinition.js";
 /**
@@ -20,7 +21,7 @@ export declare class Tool<NAME extends string, INPUT, OUTPUT> {
      * The schema of the input that the tool expects. The language model will use this to generate the input.
      * Use descriptions to make the input understandable for the language model.
      */
-    readonly inputSchema: Schema<INPUT>;
+    readonly inputSchema: Schema<INPUT> & JsonSchemaProducer;
     /**
      * An optional schema of the output that the tool produces. This will be used to validate the output.
      */
@@ -32,7 +33,7 @@ export declare class Tool<NAME extends string, INPUT, OUTPUT> {
     constructor({ name, description, inputSchema, outputSchema, execute, }: {
         name: NAME;
         description: string;
-        inputSchema: Schema<INPUT>;
+        inputSchema: Schema<INPUT> & JsonSchemaProducer;
         outputSchema?: Schema<OUTPUT>;
         execute(input: INPUT, options?: FunctionOptions): PromiseLike<OUTPUT>;
     });

package/vector-index/VectorIndexRetriever.cjs

@@ -1,7 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.VectorIndexRetriever = void 0;
-const embedText_js_1 = require("../model-function/embed-text/embedText.cjs");
+const embed_js_1 = require("../model-function/embed/embed.cjs");
 class VectorIndexRetriever {
     constructor({ vectorIndex, embeddingModel, maxResults, similarityThreshold, filter, }) {
         Object.defineProperty(this, "vectorIndex", {
@@ -38,12 +38,7 @@ class VectorIndexRetriever {
                 run: options.run,
             });
         }
-        let filter = this.settings?.filter;
-        if (typeof query === "object") {
-            filter = query.filter ?? filter; // use filter from query if available
-            query = query.text;
-        }
-        const embedding = await (0, embedText_js_1.embedText)(this.embeddingModel, query, {
+        const embedding = await (0, embed_js_1.embed)(this.embeddingModel, query, {
             functionId: options?.functionId,
             run: options?.run,
         });
@@ -51,7 +46,7 @@ class VectorIndexRetriever {
             queryVector: embedding,
             maxResults: this.settings.maxResults ?? 1,
             similarityThreshold: this.settings.similarityThreshold,
-            filter,
+            filter: this.settings?.filter,
         });
         return queryResult.map((item) => item.data);
     }

package/vector-index/VectorIndexRetriever.d.ts

@@ -1,5 +1,5 @@
+import { EmbeddingModel, EmbeddingModelSettings } from "../model-function/embed/EmbeddingModel.js";
 import { ModelFunctionOptions } from "../model-function/ModelFunctionOptions.js";
-import { TextEmbeddingModel, TextEmbeddingModelSettings } from "../model-function/embed-text/TextEmbeddingModel.js";
 import { Retriever, RetrieverSettings } from "../retriever/Retriever.js";
 import { VectorIndex } from "./VectorIndex.js";
 export interface VectorIndexRetrieverSettings<FILTER> {
@@ -7,19 +7,14 @@ export interface VectorIndexRetrieverSettings<FILTER> {
     similarityThreshold?: number;
     filter?: FILTER;
 }
-type VectorIndexRetrieverQuery<FILTER> = string | {
-    text: string;
-    filter?: FILTER;
-};
-export declare class VectorIndexRetriever<OBJECT, INDEX, FILTER> implements Retriever<OBJECT, VectorIndexRetrieverQuery<FILTER>, VectorIndexRetrieverSettings<FILTER>> {
+export declare class VectorIndexRetriever<OBJECT, VALUE, INDEX, FILTER> implements Retriever<OBJECT, VALUE, VectorIndexRetrieverSettings<FILTER>> {
     private readonly vectorIndex;
     private readonly embeddingModel;
     private readonly settings;
     constructor({ vectorIndex, embeddingModel, maxResults, similarityThreshold, filter, }: {
         vectorIndex: VectorIndex<OBJECT, INDEX, FILTER>;
-        embeddingModel: TextEmbeddingModel<unknown, TextEmbeddingModelSettings>;
+        embeddingModel: EmbeddingModel<VALUE, unknown, EmbeddingModelSettings>;
     } & VectorIndexRetrieverSettings<FILTER>);
-    retrieve(query: VectorIndexRetrieverQuery<FILTER>, options?: ModelFunctionOptions<RetrieverSettings>): Promise<OBJECT[]>;
+    retrieve(query: VALUE, options?: ModelFunctionOptions<RetrieverSettings>): Promise<OBJECT[]>;
     withSettings(additionalSettings: Partial<VectorIndexRetrieverSettings<FILTER>>): this;
 }
-export {};

package/vector-index/VectorIndexRetriever.js

@@ -1,4 +1,4 @@
-import { embedText } from "../model-function/embed-text/embedText.js";
+import { embed } from "../model-function/embed/embed.js";
 export class VectorIndexRetriever {
     constructor({ vectorIndex, embeddingModel, maxResults, similarityThreshold, filter, }) {
         Object.defineProperty(this, "vectorIndex", {
@@ -35,12 +35,7 @@ export class VectorIndexRetriever {
                 run: options.run,
             });
         }
-        let filter = this.settings?.filter;
-        if (typeof query === "object") {
-            filter = query.filter ?? filter; // use filter from query if available
-            query = query.text;
-        }
-        const embedding = await embedText(this.embeddingModel, query, {
+        const embedding = await embed(this.embeddingModel, query, {
             functionId: options?.functionId,
             run: options?.run,
         });
@@ -48,7 +43,7 @@ export class VectorIndexRetriever {
             queryVector: embedding,
             maxResults: this.settings.maxResults ?? 1,
             similarityThreshold: this.settings.similarityThreshold,
-            filter,
+            filter: this.settings?.filter,
         });
         return queryResult.map((item) => item.data);
     }

package/vector-index/memory/MemoryVectorIndex.cjs

@@ -4,8 +4,14 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 };
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.MemoryVectorIndex = void 0;
+const secure_json_parse_1 = __importDefault(require("secure-json-parse"));
 const zod_1 = __importDefault(require("zod"));
 const cosineSimilarity_js_1 = require("../../util/cosineSimilarity.cjs");
+const jsonDataSchema = zod_1.default.array(zod_1.default.object({
+    id: zod_1.default.string(),
+    vector: zod_1.default.array(zod_1.default.number()),
+    data: zod_1.default.unknown(),
+}));
 /**
  * A very simple vector index that stores all entries in memory. Useful when you only have
  * a small number of entries and don't want to set up a real database, e.g. for conversational memory
@@ -21,15 +27,16 @@ class MemoryVectorIndex {
         });
     }
     static async deserialize({ serializedData, schema, }) {
-        let json = JSON.parse(serializedData);
+        // validate the outer structure:
+        const json = jsonDataSchema.parse(secure_json_parse_1.default.parse(serializedData));
         if (schema != null) {
-            json = zod_1.default
-                .array(zod_1.default.object({
-                id: zod_1.default.string(),
-                vector: zod_1.default.array(zod_1.default.number()),
-                data: schema,
-            }))
-                .parse(json);
+            // when a schema is provided, validate all entries:
+            for (const entry of json) {
+                const validationResult = schema.validate(entry.data);
+                if (!validationResult.success) {
+                    throw validationResult.error;
+                }
+            }
         }
         const vectorIndex = new MemoryVectorIndex();
         vectorIndex.upsertMany(json);

package/vector-index/memory/MemoryVectorIndex.d.ts

@@ -1,5 +1,5 @@
-import z from "zod";
 import { Vector } from "../../core/Vector.js";
+import { Schema } from "../../core/structure/Schema.js";
 import { VectorIndex } from "../VectorIndex.js";
 /**
  * A very simple vector index that stores all entries in memory. Useful when you only have
@@ -9,7 +9,7 @@ import { VectorIndex } from "../VectorIndex.js";
 export declare class MemoryVectorIndex<DATA> implements VectorIndex<DATA, MemoryVectorIndex<DATA>, (value: DATA) => boolean> {
     static deserialize<DATA>({ serializedData, schema, }: {
         serializedData: string;
-        schema?: z.ZodSchema<DATA>;
+        schema?: Schema<DATA>;
     }): Promise<MemoryVectorIndex<DATA>>;
     private readonly entries;
     upsertMany(data: Array<{