qwen-embedder 0.1.1 → 0.1.4

package/dist/index.d.ts

@@ -4,12 +4,23 @@ interface EmbedderOptions {
     cacheSize?: number;
 }
 
+interface SyncEmbedderOptions {
+    cacheSize?: number;
+}
+declare class SyncEmbedder {
+    private fn;
+    constructor(options?: SyncEmbedderOptions);
+    embedSync(text: string): number[];
+    embedSync(texts: string[]): number[][];
+}
+
 declare class Embedder {
     private pipe;
     private cache;
     private maxCacheSize;
     private queue;
     private constructor();
+    static createSync(options?: SyncEmbedderOptions): SyncEmbedder;
     static create(options?: EmbedderOptions): Promise<Embedder>;
     embed(text: string): Promise<number[]>;
     embed(texts: string[]): Promise<number[][]>;
@@ -19,4 +30,4 @@ declare class Embedder {
     private setCache;
 }
 
-export { Embedder, type EmbedderOptions };
+export { Embedder, type EmbedderOptions, SyncEmbedder, type SyncEmbedderOptions };

package/dist/index.js

@@ -3,6 +3,28 @@ import { pipeline } from "@huggingface/transformers";
 import PQueue from "p-queue";
 import { homedir } from "os";
 import { join } from "path";
+
+// src/sync-embedder.ts
+import { createSyncFn } from "synckit";
+import { fileURLToPath } from "url";
+import { dirname, resolve } from "path";
+var __dirname = dirname(fileURLToPath(import.meta.url));
+var workerPath = resolve(__dirname, "sync-worker.js");
+var SyncEmbedder = class {
+  fn;
+  constructor(options = {}) {
+    this.fn = createSyncFn(workerPath);
+    this.fn({ _cmd: "init", cacheSize: options.cacheSize ?? 100 });
+  }
+  embedSync(input) {
+    if (Array.isArray(input)) {
+      return this.fn({ _cmd: "embed", texts: input });
+    }
+    return this.fn({ _cmd: "embed", text: input });
+  }
+};
+
+// src/embedder.ts
 var MODEL_ID = "onnx-community/Qwen3-Embedding-0.6B-ONNX";
 var CACHE_DIR = join(homedir(), ".qwenembedder", ".cache", "models");
 function resolveConcurrency(options) {
@@ -21,6 +43,9 @@ var Embedder = class _Embedder {
     this.cache = /* @__PURE__ */ new Map();
     this.queue = concurrency !== null ? new PQueue({ concurrency }) : null;
   }
+  static createSync(options = {}) {
+    return new SyncEmbedder(options);
+  }
   static async create(options = {}) {
     const cacheSize = options.cacheSize ?? 100;
     const concurrency = resolveConcurrency(options);
@@ -53,7 +78,8 @@ var Embedder = class _Embedder {
       return [];
     }
     const results = new Array(texts.length);
-    const uncached = [];
+    const uncachedByText = /* @__PURE__ */ new Map();
+    const uncachedTexts = [];
     for (let i = 0; i < texts.length; i++) {
       const t = texts[i];
       if (typeof t !== "string" || t.length === 0) {
@@ -61,25 +87,35 @@ var Embedder = class _Embedder {
       }
       const cached = this.cache.get(t);
       if (cached !== void 0) {
-        results[i] = cached;
+        results[i] = cached.slice();
       } else {
-        uncached.push(i);
+        const existing = uncachedByText.get(t);
+        if (existing !== void 0) {
+          existing.push(i);
+        } else {
+          uncachedByText.set(t, [i]);
+          uncachedTexts.push(t);
+        }
       }
     }
-    if (uncached.length === 0) {
+    if (uncachedTexts.length === 0) {
       return results;
     }
-    const uncachedTexts = uncached.map((i) => texts[i]);
     const inferred = await this.runInference(uncachedTexts);
-    for (let j = 0; j < uncached.length; j++) {
+    for (let j = 0; j < uncachedTexts.length; j++) {
+      const text = uncachedTexts[j];
       const vector = inferred[j];
-      this.setCache(uncachedTexts[j], vector);
-      results[uncached[j]] = vector;
+      const stored = vector.slice();
+      this.setCache(text, stored);
+      const positions = uncachedByText.get(text) ?? [];
+      for (const position of positions) {
+        results[position] = stored.slice();
+      }
     }
     return results;
   }
   async runInference(input) {
-    const exec = () => this.pipe(input, { pooling: "mean", normalize: true });
+    const exec = () => this.pipe(input, { pooling: "last_token", normalize: true });
     const output = this.queue ? await this.queue.add(exec) : await exec();
     const data = Array.from(output.data);
     const hiddenDim = output.dims[output.dims.length - 1];
@@ -94,16 +130,18 @@ var Embedder = class _Embedder {
   }
   setCache(key, value) {
     if (this.maxCacheSize === 0) return;
+    const stored = value.slice();
     if (this.cache.size >= this.maxCacheSize) {
       const oldest = this.cache.keys().next();
       if (!oldest.done) {
         this.cache.delete(oldest.value);
       }
     }
-    this.cache.set(key, value);
+    this.cache.set(key, stored);
   }
 };
 export {
-  Embedder
+  Embedder,
+  SyncEmbedder
 };
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/embedder.ts"],"sourcesContent":["import { pipeline } from '@huggingface/transformers'\nimport PQueue from 'p-queue'\nimport { homedir } from 'os'\nimport { join } from 'path'\nimport type { EmbedderOptions } from './types.js'\n\nconst MODEL_ID = 'onnx-community/Qwen3-Embedding-0.6B-ONNX'\nconst CACHE_DIR = join(homedir(), '.qwenembedder', '.cache', 'models')\n\nfunction resolveConcurrency(options: EmbedderOptions): number | null {\n  if (options.concurrency !== undefined) return options.concurrency\n  if (options.queue) return 1\n  return null\n}\n\ninterface InferenceResult {\n  data: Float32Array | number[]\n  dims: number[]\n  tolist(): number[][]\n}\n\nexport class Embedder {\n  private pipe: (texts: string | string[], options?: Record<string, unknown>) => Promise<InferenceResult>\n  private cache: Map<string, number[]>\n  private maxCacheSize: number\n  private queue: PQueue | null\n\n  private constructor(\n    pipe: Embedder['pipe'],\n    concurrency: number | null,\n    maxCacheSize: number,\n  ) {\n    this.pipe = pipe\n    this.maxCacheSize = maxCacheSize\n    this.cache = new Map()\n    this.queue = concurrency !== null ? new PQueue({ concurrency }) : null\n  }\n\n  static async create(options: EmbedderOptions = {}): Promise<Embedder> {\n    const cacheSize = options.cacheSize ?? 100\n    const concurrency = resolveConcurrency(options)\n\n    const pipe = await pipeline('feature-extraction', MODEL_ID, {\n      cache_dir: CACHE_DIR,\n      dtype: 'q8',\n    })\n\n    return new Embedder(pipe as unknown as Embedder['pipe'], concurrency, cacheSize)\n  }\n\n  async embed(text: string): Promise<number[]>\n  async embed(texts: string[]): Promise<number[][]>\n  async embed(input: string | string[]): Promise<number[] | number[][]> {\n    if (Array.isArray(input)) {\n      return this.embedBatch(input)\n    }\n    return this.embedSingle(input)\n  }\n\n  private async embedSingle(text: string): Promise<number[]> {\n    if (text.length === 0) {\n      throw new TypeError('Input must be a non-empty string')\n    }\n\n    const cached = this.cache.get(text)\n    if (cached !== undefined) {\n      return cached.slice()\n    }\n\n    const result = await this.runInference(text)\n    this.setCache(text, result)\n    return result.slice()\n  }\n\n  private async embedBatch(texts: string[]): Promise<number[][]> {\n    if (texts.length === 0) {\n      return []\n    }\n\n    const results: (number[] | undefined)[] = new Array(texts.length)\n    const uncached: number[] = []\n\n    for (let i = 0; i < texts.length; i++) {\n      const t = texts[i]\n      if (typeof t !== 'string' || t.length === 0) {\n        throw new TypeError('Input must be a non-empty string')\n      }\n      const cached = this.cache.get(t)\n      if (cached !== undefined) {\n        results[i] = cached\n      } else {\n        uncached.push(i)\n      }\n    }\n\n    if (uncached.length === 0) {\n      return results as number[][]\n    }\n\n    const uncachedTexts = uncached.map(i => texts[i])\n    const inferred = await this.runInference(uncachedTexts)\n\n    for (let j = 0; j < uncached.length; j++) {\n      const vector = inferred[j]\n      this.setCache(uncachedTexts[j], vector)\n      results[uncached[j]] = vector\n    }\n\n    return results as number[][]\n  }\n\n  private async runInference(text: string): Promise<number[]>\n  private async runInference(texts: string[]): Promise<number[][]>\n  private async runInference(input: string | string[]): Promise<number[] | number[][]> {\n    const exec = () => this.pipe(input, { pooling: 'mean', normalize: true })\n\n    const output = this.queue\n      ? await this.queue.add(exec)\n      : await exec()\n\n    const data = Array.from(output.data) as number[]\n    const hiddenDim = output.dims[output.dims.length - 1]\n\n    if (typeof input === 'string') {\n      return data\n    }\n\n    const result: number[][] = []\n    for (let i = 0; i < data.length; i += hiddenDim) {\n      result.push(data.slice(i, i + hiddenDim))\n    }\n    return result\n  }\n\n  private setCache(key: string, value: number[]): void {\n    if (this.maxCacheSize === 0) return\n\n    if (this.cache.size >= this.maxCacheSize) {\n      const oldest = this.cache.keys().next()\n      if (!oldest.done) {\n        this.cache.delete(oldest.value)\n      }\n    }\n    this.cache.set(key, value)\n  }\n}\n"],"mappings":";AAAA,SAAS,gBAAgB;AACzB,OAAO,YAAY;AACnB,SAAS,eAAe;AACxB,SAAS,YAAY;AAGrB,IAAM,WAAW;AACjB,IAAM,YAAY,KAAK,QAAQ,GAAG,iBAAiB,UAAU,QAAQ;AAErE,SAAS,mBAAmB,SAAyC;AACnE,MAAI,QAAQ,gBAAgB,OAAW,QAAO,QAAQ;AACtD,MAAI,QAAQ,MAAO,QAAO;AAC1B,SAAO;AACT;AAQO,IAAM,WAAN,MAAM,UAAS;AAAA,EACZ;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACN,MACA,aACA,cACA;AACA,SAAK,OAAO;AACZ,SAAK,eAAe;AACpB,SAAK,QAAQ,oBAAI,IAAI;AACrB,SAAK,QAAQ,gBAAgB,OAAO,IAAI,OAAO,EAAE,YAAY,CAAC,IAAI;AAAA,EACpE;AAAA,EAEA,aAAa,OAAO,UAA2B,CAAC,GAAsB;AACpE,UAAM,YAAY,QAAQ,aAAa;AACvC,UAAM,cAAc,mBAAmB,OAAO;AAE9C,UAAM,OAAO,MAAM,SAAS,sBAAsB,UAAU;AAAA,MAC1D,WAAW;AAAA,MACX,OAAO;AAAA,IACT,CAAC;AAED,WAAO,IAAI,UAAS,MAAqC,aAAa,SAAS;AAAA,EACjF;AAAA,EAIA,MAAM,MAAM,OAA0D;AACpE,QAAI,MAAM,QAAQ,KAAK,GAAG;AACxB,aAAO,KAAK,WAAW,KAAK;AAAA,IAC9B;AACA,WAAO,KAAK,YAAY,KAAK;AAAA,EAC/B;AAAA,EAEA,MAAc,YAAY,MAAiC;AACzD,QAAI,KAAK,WAAW,GAAG;AACrB,YAAM,IAAI,UAAU,kCAAkC;AAAA,IACxD;AAEA,UAAM,SAAS,KAAK,MAAM,IAAI,IAAI;AAClC,QAAI,WAAW,QAAW;AACxB,aAAO,OAAO,MAAM;AAAA,IACtB;AAEA,UAAM,SAAS,MAAM,KAAK,aAAa,IAAI;AAC3C,SAAK,SAAS,MAAM,MAAM;AAC1B,WAAO,OAAO,MAAM;AAAA,EACtB;AAAA,EAEA,MAAc,WAAW,OAAsC;AAC7D,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,UAAoC,IAAI,MAAM,MAAM,MAAM;AAChE,UAAM,WAAqB,CAAC;AAE5B,aAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;AACrC,YAAM,IAAI,MAAM,CAAC;AACjB,UAAI,OAAO,MAAM,YAAY,EAAE,WAAW,GAAG;AAC3C,cAAM,IAAI,UAAU,kCAAkC;AAAA,MACxD;AACA,YAAM,SAAS,KAAK,MAAM,IAAI,CAAC;AAC/B,UAAI,WAAW,QAAW;AACxB,gBAAQ,CAAC,IAAI;AAAA,MACf,OAAO;AACL,iBAAS,KAAK,CAAC;AAAA,MACjB;AAAA,IACF;AAEA,QAAI,SAAS,WAAW,GAAG;AACzB,aAAO;AAAA,IACT;AAEA,UAAM,gBAAgB,SAAS,IAAI,OAAK,MAAM,CAAC,CAAC;AAChD,UAAM,WAAW,MAAM,KAAK,aAAa,aAAa;AAEtD,aAAS,IAAI,GAAG,IAAI,SAAS,QAAQ,KAAK;AACxC,YAAM,SAAS,SAAS,CAAC;AACzB,WAAK,SAAS,cAAc,CAAC,GAAG,MAAM;AACtC,cAAQ,SAAS,CAAC,CAAC,IAAI;AAAA,IACzB;AAEA,WAAO;AAAA,EACT;AAAA,EAIA,MAAc,aAAa,OAA0D;AACnF,UAAM,OAAO,MAAM,KAAK,KAAK,OAAO,EAAE,SAAS,QAAQ,WAAW,KAAK,CAAC;AAExE,UAAM,SAAS,KAAK,QAChB,MAAM,KAAK,MAAM,IAAI,IAAI,IACzB,MAAM,KAAK;AAEf,UAAM,OAAO,MAAM,KAAK,OAAO,IAAI;AACnC,UAAM,YAAY,OAAO,KAAK,OAAO,KAAK,SAAS,CAAC;AAEpD,QAAI,OAAO,UAAU,UAAU;AAC7B,aAAO;AAAA,IACT;AAEA,UAAM,SAAqB,CAAC;AAC5B,aAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK,WAAW;AAC/C,aAAO,KAAK,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC;AAAA,IAC1C;AACA,WAAO;AAAA,EACT;AAAA,EAEQ,SAAS,KAAa,OAAuB;AACnD,QAAI,KAAK,iBAAiB,EAAG;AAE7B,QAAI,KAAK,MAAM,QAAQ,KAAK,cAAc;AACxC,YAAM,SAAS,KAAK,MAAM,KAAK,EAAE,KAAK;AACtC,UAAI,CAAC,OAAO,MAAM;AAChB,aAAK,MAAM,OAAO,OAAO,KAAK;AAAA,MAChC;AAAA,IACF;AACA,SAAK,MAAM,IAAI,KAAK,KAAK;AAAA,EAC3B;AACF;","names":[]}
+{"version":3,"sources":["../src/embedder.ts","../src/sync-embedder.ts"],"sourcesContent":["import { pipeline } from '@huggingface/transformers'\nimport PQueue from 'p-queue'\nimport { homedir } from 'os'\nimport { join } from 'path'\nimport type { EmbedderOptions } from './types.js'\nimport { SyncEmbedder } from './sync-embedder.js'\nimport type { SyncEmbedderOptions } from './sync-embedder.js'\n\nconst MODEL_ID = 'onnx-community/Qwen3-Embedding-0.6B-ONNX'\nconst CACHE_DIR = join(homedir(), '.qwenembedder', '.cache', 'models')\n\nfunction resolveConcurrency(options: EmbedderOptions): number | null {\n  if (options.concurrency !== undefined) return options.concurrency\n  if (options.queue) return 1\n  return null\n}\n\ninterface InferenceResult {\n  data: Float32Array | number[]\n  dims: number[]\n  tolist(): number[][]\n}\n\nexport class Embedder {\n  private pipe: (texts: string | string[], options?: Record<string, unknown>) => Promise<InferenceResult>\n  private cache: Map<string, number[]>\n  private maxCacheSize: number\n  private queue: PQueue | null\n\n  private constructor(\n    pipe: Embedder['pipe'],\n    concurrency: number | null,\n    maxCacheSize: number,\n  ) {\n    this.pipe = pipe\n    this.maxCacheSize = maxCacheSize\n    this.cache = new Map()\n    this.queue = concurrency !== null ? new PQueue({ concurrency }) : null\n  }\n\n  static createSync(options: SyncEmbedderOptions = {}): SyncEmbedder {\n    return new SyncEmbedder(options)\n  }\n\n  static async create(options: EmbedderOptions = {}): Promise<Embedder> {\n    const cacheSize = options.cacheSize ?? 100\n    const concurrency = resolveConcurrency(options)\n\n    const pipe = await pipeline('feature-extraction', MODEL_ID, {\n      cache_dir: CACHE_DIR,\n      dtype: 'q8',\n    })\n\n    return new Embedder(pipe as unknown as Embedder['pipe'], concurrency, cacheSize)\n  }\n\n  async embed(text: string): Promise<number[]>\n  async embed(texts: string[]): Promise<number[][]>\n  async embed(input: string | string[]): Promise<number[] | number[][]> {\n    if (Array.isArray(input)) {\n      return this.embedBatch(input)\n    }\n    return this.embedSingle(input)\n  }\n\n  private async embedSingle(text: string): Promise<number[]> {\n    if (text.length === 0) {\n      throw new TypeError('Input must be a non-empty string')\n    }\n\n    const cached = this.cache.get(text)\n    if (cached !== undefined) {\n      return cached.slice()\n    }\n\n    const result = await this.runInference(text)\n    this.setCache(text, result)\n    return result.slice()\n  }\n\n  private async embedBatch(texts: string[]): Promise<number[][]> {\n    if (texts.length === 0) {\n      return []\n    }\n\n    const results: (number[] | undefined)[] = new Array(texts.length)\n    const uncachedByText = new Map<string, number[]>()\n    const uncachedTexts: string[] = []\n\n    for (let i = 0; i < texts.length; i++) {\n      const t = texts[i]\n      if (typeof t !== 'string' || t.length === 0) {\n        throw new TypeError('Input must be a non-empty string')\n      }\n      const cached = this.cache.get(t)\n      if (cached !== undefined) {\n        results[i] = cached.slice()\n      } else {\n        const existing = uncachedByText.get(t)\n        if (existing !== undefined) {\n          existing.push(i)\n        } else {\n          uncachedByText.set(t, [i])\n          uncachedTexts.push(t)\n        }\n      }\n    }\n\n    if (uncachedTexts.length === 0) {\n      return results as number[][]\n    }\n\n    const inferred = await this.runInference(uncachedTexts)\n\n    for (let j = 0; j < uncachedTexts.length; j++) {\n      const text = uncachedTexts[j]\n      const vector = inferred[j]\n      const stored = vector.slice()\n      this.setCache(text, stored)\n      const positions = uncachedByText.get(text) ?? []\n      for (const position of positions) {\n        results[position] = stored.slice()\n      }\n    }\n\n    return results as number[][]\n  }\n\n  private async runInference(text: string): Promise<number[]>\n  private async runInference(texts: string[]): Promise<number[][]>\n  private async runInference(input: string | string[]): Promise<number[] | number[][]> {\n    const exec = () => this.pipe(input, { pooling: 'last_token', normalize: true })\n\n    const output = this.queue\n      ? await this.queue.add(exec)\n      : await exec()\n\n    const data = Array.from(output.data) as number[]\n    const hiddenDim = output.dims[output.dims.length - 1]\n\n    if (typeof input === 'string') {\n      return data\n    }\n\n    const result: number[][] = []\n    for (let i = 0; i < data.length; i += hiddenDim) {\n      result.push(data.slice(i, i + hiddenDim))\n    }\n    return result\n  }\n\n  private setCache(key: string, value: number[]): void {\n    if (this.maxCacheSize === 0) return\n\n    const stored = value.slice()\n    if (this.cache.size >= this.maxCacheSize) {\n      const oldest = this.cache.keys().next()\n      if (!oldest.done) {\n        this.cache.delete(oldest.value)\n      }\n    }\n    this.cache.set(key, stored)\n  }\n}\n","import { createSyncFn } from 'synckit'\nimport { fileURLToPath } from 'url'\nimport { dirname, resolve } from 'path'\n\nconst __dirname = dirname(fileURLToPath(import.meta.url))\nconst workerPath = resolve(__dirname, 'sync-worker.js')\n\ntype SyncFn = {\n  (command: { _cmd: 'init'; cacheSize?: number }): void\n  (command: { _cmd: 'embed'; text: string }): number[]\n  (command: { _cmd: 'embed'; texts: string[] }): number[][]\n}\n\nexport interface SyncEmbedderOptions {\n  cacheSize?: number\n}\n\nexport class SyncEmbedder {\n  private fn: SyncFn\n\n  constructor(options: SyncEmbedderOptions = {}) {\n    this.fn = createSyncFn(workerPath) as SyncFn\n    this.fn({ _cmd: 'init', cacheSize: options.cacheSize ?? 100 })\n  }\n\n  embedSync(text: string): number[]\n  embedSync(texts: string[]): number[][]\n  embedSync(input: string | string[]): number[] | number[][] {\n    if (Array.isArray(input)) {\n      return this.fn({ _cmd: 'embed', texts: input })\n    }\n    return this.fn({ _cmd: 'embed', text: input })\n  }\n}\n"],"mappings":";AAAA,SAAS,gBAAgB;AACzB,OAAO,YAAY;AACnB,SAAS,eAAe;AACxB,SAAS,YAAY;;;ACHrB,SAAS,oBAAoB;AAC7B,SAAS,qBAAqB;AAC9B,SAAS,SAAS,eAAe;AAEjC,IAAM,YAAY,QAAQ,cAAc,YAAY,GAAG,CAAC;AACxD,IAAM,aAAa,QAAQ,WAAW,gBAAgB;AAY/C,IAAM,eAAN,MAAmB;AAAA,EAChB;AAAA,EAER,YAAY,UAA+B,CAAC,GAAG;AAC7C,SAAK,KAAK,aAAa,UAAU;AACjC,SAAK,GAAG,EAAE,MAAM,QAAQ,WAAW,QAAQ,aAAa,IAAI,CAAC;AAAA,EAC/D;AAAA,EAIA,UAAU,OAAiD;AACzD,QAAI,MAAM,QAAQ,KAAK,GAAG;AACxB,aAAO,KAAK,GAAG,EAAE,MAAM,SAAS,OAAO,MAAM,CAAC;AAAA,IAChD;AACA,WAAO,KAAK,GAAG,EAAE,MAAM,SAAS,MAAM,MAAM,CAAC;AAAA,EAC/C;AACF;;;ADzBA,IAAM,WAAW;AACjB,IAAM,YAAY,KAAK,QAAQ,GAAG,iBAAiB,UAAU,QAAQ;AAErE,SAAS,mBAAmB,SAAyC;AACnE,MAAI,QAAQ,gBAAgB,OAAW,QAAO,QAAQ;AACtD,MAAI,QAAQ,MAAO,QAAO;AAC1B,SAAO;AACT;AAQO,IAAM,WAAN,MAAM,UAAS;AAAA,EACZ;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACN,MACA,aACA,cACA;AACA,SAAK,OAAO;AACZ,SAAK,eAAe;AACpB,SAAK,QAAQ,oBAAI,IAAI;AACrB,SAAK,QAAQ,gBAAgB,OAAO,IAAI,OAAO,EAAE,YAAY,CAAC,IAAI;AAAA,EACpE;AAAA,EAEA,OAAO,WAAW,UAA+B,CAAC,GAAiB;AACjE,WAAO,IAAI,aAAa,OAAO;AAAA,EACjC;AAAA,EAEA,aAAa,OAAO,UAA2B,CAAC,GAAsB;AACpE,UAAM,YAAY,QAAQ,aAAa;AACvC,UAAM,cAAc,mBAAmB,OAAO;AAE9C,UAAM,OAAO,MAAM,SAAS,sBAAsB,UAAU;AAAA,MAC1D,WAAW;AAAA,MACX,OAAO;AAAA,IACT,CAAC;AAED,WAAO,IAAI,UAAS,MAAqC,aAAa,SAAS;AAAA,EACjF;AAAA,EAIA,MAAM,MAAM,OAA0D;AACpE,QAAI,MAAM,QAAQ,KAAK,GAAG;AACxB,aAAO,KAAK,WAAW,KAAK;AAAA,IAC9B;AACA,WAAO,KAAK,YAAY,KAAK;AAAA,EAC/B;AAAA,EAEA,MAAc,YAAY,MAAiC;AACzD,QAAI,KAAK,WAAW,GAAG;AACrB,YAAM,IAAI,UAAU,kCAAkC;AAAA,IACxD;AAEA,UAAM,SAAS,KAAK,MAAM,IAAI,IAAI;AAClC,QAAI,WAAW,QAAW;AACxB,aAAO,OAAO,MAAM;AAAA,IACtB;AAEA,UAAM,SAAS,MAAM,KAAK,aAAa,IAAI;AAC3C,SAAK,SAAS,MAAM,MAAM;AAC1B,WAAO,OAAO,MAAM;AAAA,EACtB;AAAA,EAEA,MAAc,WAAW,OAAsC;AAC7D,QAAI,MAAM,WAAW,GAAG;AACtB,aAAO,CAAC;AAAA,IACV;AAEA,UAAM,UAAoC,IAAI,MAAM,MAAM,MAAM;AAChE,UAAM,iBAAiB,oBAAI,IAAsB;AACjD,UAAM,gBAA0B,CAAC;AAEjC,aAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;AACrC,YAAM,IAAI,MAAM,CAAC;AACjB,UAAI,OAAO,MAAM,YAAY,EAAE,WAAW,GAAG;AAC3C,cAAM,IAAI,UAAU,kCAAkC;AAAA,MACxD;AACA,YAAM,SAAS,KAAK,MAAM,IAAI,CAAC;AAC/B,UAAI,WAAW,QAAW;AACxB,gBAAQ,CAAC,IAAI,OAAO,MAAM;AAAA,MAC5B,OAAO;AACL,cAAM,WAAW,eAAe,IAAI,CAAC;AACrC,YAAI,aAAa,QAAW;AAC1B,mBAAS,KAAK,CAAC;AAAA,QACjB,OAAO;AACL,yBAAe,IAAI,GAAG,CAAC,CAAC,CAAC;AACzB,wBAAc,KAAK,CAAC;AAAA,QACtB;AAAA,MACF;AAAA,IACF;AAEA,QAAI,cAAc,WAAW,GAAG;AAC9B,aAAO;AAAA,IACT;AAEA,UAAM,WAAW,MAAM,KAAK,aAAa,aAAa;AAEtD,aAAS,IAAI,GAAG,IAAI,cAAc,QAAQ,KAAK;AAC7C,YAAM,OAAO,cAAc,CAAC;AAC5B,YAAM,SAAS,SAAS,CAAC;AACzB,YAAM,SAAS,OAAO,MAAM;AAC5B,WAAK,SAAS,MAAM,MAAM;AAC1B,YAAM,YAAY,eAAe,IAAI,IAAI,KAAK,CAAC;AAC/C,iBAAW,YAAY,WAAW;AAChC,gBAAQ,QAAQ,IAAI,OAAO,MAAM;AAAA,MACnC;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA,EAIA,MAAc,aAAa,OAA0D;AACnF,UAAM,OAAO,MAAM,KAAK,KAAK,OAAO,EAAE,SAAS,cAAc,WAAW,KAAK,CAAC;AAE9E,UAAM,SAAS,KAAK,QAChB,MAAM,KAAK,MAAM,IAAI,IAAI,IACzB,MAAM,KAAK;AAEf,UAAM,OAAO,MAAM,KAAK,OAAO,IAAI;AACnC,UAAM,YAAY,OAAO,KAAK,OAAO,KAAK,SAAS,CAAC;AAEpD,QAAI,OAAO,UAAU,UAAU;AAC7B,aAAO;AAAA,IACT;AAEA,UAAM,SAAqB,CAAC;AAC5B,aAAS,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK,WAAW;AAC/C,aAAO,KAAK,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC;AAAA,IAC1C;AACA,WAAO;AAAA,EACT;AAAA,EAEQ,SAAS,KAAa,OAAuB;AACnD,QAAI,KAAK,iBAAiB,EAAG;AAE7B,UAAM,SAAS,MAAM,MAAM;AAC3B,QAAI,KAAK,MAAM,QAAQ,KAAK,cAAc;AACxC,YAAM,SAAS,KAAK,MAAM,KAAK,EAAE,KAAK;AACtC,UAAI,CAAC,OAAO,MAAM;AAChB,aAAK,MAAM,OAAO,OAAO,KAAK;AAAA,MAChC;AAAA,IACF;AACA,SAAK,MAAM,IAAI,KAAK,MAAM;AAAA,EAC5B;AACF;","names":[]}

package/dist/sync-worker.d.ts

@@ -0,0 +1,2 @@
+
+export {  }

package/dist/sync-worker.js

@@ -0,0 +1,81 @@
+// src/sync-worker.ts
+import { pipeline } from "@huggingface/transformers";
+import { runAsWorker } from "synckit";
+import { homedir } from "os";
+import { join } from "path";
+var MODEL_ID = "onnx-community/Qwen3-Embedding-0.6B-ONNX";
+var CACHE_DIR = join(homedir(), ".qwenembedder", ".cache", "models");
+var pipe = null;
+var cache = null;
+var maxCacheSize = 100;
+function setCache(key, value) {
+  if (!cache) return;
+  if (maxCacheSize === 0) return;
+  if (cache.size >= maxCacheSize) {
+    const oldest = cache.keys().next();
+    if (!oldest.done) {
+      cache.delete(oldest.value);
+    }
+  }
+  cache.set(key, value);
+}
+runAsWorker(async (command) => {
+  if (command._cmd === "init") {
+    maxCacheSize = command.cacheSize ?? 100;
+    cache = /* @__PURE__ */ new Map();
+    return;
+  }
+  if (!pipe) {
+    pipe = await pipeline("feature-extraction", MODEL_ID, {
+      cache_dir: CACHE_DIR,
+      dtype: "q8"
+    });
+  }
+  if (command._cmd === "embed") {
+    const texts = command.texts ?? (command.text ? [command.text] : []);
+    if (texts.length === 0) {
+      throw new TypeError("Input must be a non-empty string");
+    }
+    const results = new Array(texts.length);
+    const uncachedByText = /* @__PURE__ */ new Map();
+    const uncachedTexts = [];
+    for (let i = 0; i < texts.length; i++) {
+      const t = texts[i];
+      if (typeof t !== "string" || t.length === 0) {
+        throw new TypeError("Input must be a non-empty string");
+      }
+      const cached = cache?.get(t);
+      if (cached !== void 0) {
+        results[i] = cached.slice();
+      } else {
+        const existing = uncachedByText.get(t);
+        if (existing !== void 0) {
+          existing.push(i);
+        } else {
+          uncachedByText.set(t, [i]);
+          uncachedTexts.push(t);
+        }
+      }
+    }
+    if (uncachedTexts.length > 0) {
+      const output = await pipe(uncachedTexts, { pooling: "last_token", normalize: true });
+      const data = Array.from(output.data);
+      const hiddenDim = output.dims[output.dims.length - 1];
+      for (let j = 0; j < uncachedTexts.length; j++) {
+        const text = uncachedTexts[j];
+        const vector = data.slice(j * hiddenDim, (j + 1) * hiddenDim);
+        const stored = vector.slice();
+        setCache(text, stored);
+        const positions = uncachedByText.get(text) ?? [];
+        for (const position of positions) {
+          results[position] = stored.slice();
+        }
+      }
+    }
+    if (command.text !== void 0) {
+      return results[0];
+    }
+    return results;
+  }
+});
+//# sourceMappingURL=sync-worker.js.map

package/dist/sync-worker.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/sync-worker.ts"],"sourcesContent":["import { pipeline } from '@huggingface/transformers'\nimport { runAsWorker } from 'synckit'\nimport { homedir } from 'os'\nimport { join } from 'path'\n\ntype PipeFn = (texts: string[], options?: Record<string, unknown>) => Promise<{ data: Float32Array; dims: number[] }>\n\nconst MODEL_ID = 'onnx-community/Qwen3-Embedding-0.6B-ONNX'\nconst CACHE_DIR = join(homedir(), '.qwenembedder', '.cache', 'models')\n\nlet pipe: PipeFn | null = null\nlet cache: Map<string, number[]> | null = null\nlet maxCacheSize = 100\n\nfunction setCache(key: string, value: number[]): void {\n  if (!cache) return\n  if (maxCacheSize === 0) return\n  if (cache.size >= maxCacheSize) {\n    const oldest = cache.keys().next()\n    if (!oldest.done) {\n      cache.delete(oldest.value)\n    }\n  }\n  cache.set(key, value)\n}\n\ntype InitCommand = { _cmd: 'init'; cacheSize?: number }\ntype EmbedCommand = { _cmd: 'embed'; text?: string; texts?: string[] }\n\nrunAsWorker(async (command: InitCommand | EmbedCommand) => {\n  if (command._cmd === 'init') {\n    maxCacheSize = command.cacheSize ?? 100\n    cache = new Map()\n    return\n  }\n\n  if (!pipe) {\n    pipe = (await pipeline('feature-extraction', MODEL_ID, {\n      cache_dir: CACHE_DIR,\n      dtype: 'q8',\n    })) as unknown as PipeFn\n  }\n\n  if (command._cmd === 'embed') {\n    const texts = command.texts ?? (command.text ? [command.text] : [])\n\n    if (texts.length === 0) {\n      throw new TypeError('Input must be a non-empty string')\n    }\n\n    const results: (number[] | undefined)[] = new Array(texts.length)\n    const uncachedByText = new Map<string, number[]>()\n    const uncachedTexts: string[] = []\n\n    for (let i = 0; i < texts.length; i++) {\n      const t = texts[i]\n      if (typeof t !== 'string' || t.length === 0) {\n        throw new TypeError('Input must be a non-empty string')\n      }\n      const cached = cache?.get(t)\n      if (cached !== undefined) {\n        results[i] = cached.slice()\n      } else {\n        const existing = uncachedByText.get(t)\n        if (existing !== undefined) {\n          existing.push(i)\n        } else {\n          uncachedByText.set(t, [i])\n          uncachedTexts.push(t)\n        }\n      }\n    }\n\n    if (uncachedTexts.length > 0) {\n      const output = await pipe!(uncachedTexts, { pooling: 'last_token', normalize: true })\n      const data = Array.from(output.data) as number[]\n      const hiddenDim = output.dims[output.dims.length - 1]\n\n      for (let j = 0; j < uncachedTexts.length; j++) {\n        const text = uncachedTexts[j]\n        const vector = data.slice(j * hiddenDim, (j + 1) * hiddenDim)\n        const stored = vector.slice()\n        setCache(text, stored)\n        const positions = uncachedByText.get(text) ?? []\n        for (const position of positions) {\n          results[position] = stored.slice()\n        }\n      }\n    }\n\n    if (command.text !== undefined) {\n      return results[0] as number[]\n    }\n    return results as number[][]\n  }\n})\n"],"mappings":";AAAA,SAAS,gBAAgB;AACzB,SAAS,mBAAmB;AAC5B,SAAS,eAAe;AACxB,SAAS,YAAY;AAIrB,IAAM,WAAW;AACjB,IAAM,YAAY,KAAK,QAAQ,GAAG,iBAAiB,UAAU,QAAQ;AAErE,IAAI,OAAsB;AAC1B,IAAI,QAAsC;AAC1C,IAAI,eAAe;AAEnB,SAAS,SAAS,KAAa,OAAuB;AACpD,MAAI,CAAC,MAAO;AACZ,MAAI,iBAAiB,EAAG;AACxB,MAAI,MAAM,QAAQ,cAAc;AAC9B,UAAM,SAAS,MAAM,KAAK,EAAE,KAAK;AACjC,QAAI,CAAC,OAAO,MAAM;AAChB,YAAM,OAAO,OAAO,KAAK;AAAA,IAC3B;AAAA,EACF;AACA,QAAM,IAAI,KAAK,KAAK;AACtB;AAKA,YAAY,OAAO,YAAwC;AACzD,MAAI,QAAQ,SAAS,QAAQ;AAC3B,mBAAe,QAAQ,aAAa;AACpC,YAAQ,oBAAI,IAAI;AAChB;AAAA,EACF;AAEA,MAAI,CAAC,MAAM;AACT,WAAQ,MAAM,SAAS,sBAAsB,UAAU;AAAA,MACrD,WAAW;AAAA,MACX,OAAO;AAAA,IACT,CAAC;AAAA,EACH;AAEA,MAAI,QAAQ,SAAS,SAAS;AAC5B,UAAM,QAAQ,QAAQ,UAAU,QAAQ,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC;AAEjE,QAAI,MAAM,WAAW,GAAG;AACtB,YAAM,IAAI,UAAU,kCAAkC;AAAA,IACxD;AAEA,UAAM,UAAoC,IAAI,MAAM,MAAM,MAAM;AAChE,UAAM,iBAAiB,oBAAI,IAAsB;AACjD,UAAM,gBAA0B,CAAC;AAEjC,aAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;AACrC,YAAM,IAAI,MAAM,CAAC;AACjB,UAAI,OAAO,MAAM,YAAY,EAAE,WAAW,GAAG;AAC3C,cAAM,IAAI,UAAU,kCAAkC;AAAA,MACxD;AACA,YAAM,SAAS,OAAO,IAAI,CAAC;AAC3B,UAAI,WAAW,QAAW;AACxB,gBAAQ,CAAC,IAAI,OAAO,MAAM;AAAA,MAC5B,OAAO;AACL,cAAM,WAAW,eAAe,IAAI,CAAC;AACrC,YAAI,aAAa,QAAW;AAC1B,mBAAS,KAAK,CAAC;AAAA,QACjB,OAAO;AACL,yBAAe,IAAI,GAAG,CAAC,CAAC,CAAC;AACzB,wBAAc,KAAK,CAAC;AAAA,QACtB;AAAA,MACF;AAAA,IACF;AAEA,QAAI,cAAc,SAAS,GAAG;AAC5B,YAAM,SAAS,MAAM,KAAM,eAAe,EAAE,SAAS,cAAc,WAAW,KAAK,CAAC;AACpF,YAAM,OAAO,MAAM,KAAK,OAAO,IAAI;AACnC,YAAM,YAAY,OAAO,KAAK,OAAO,KAAK,SAAS,CAAC;AAEpD,eAAS,IAAI,GAAG,IAAI,cAAc,QAAQ,KAAK;AAC7C,cAAM,OAAO,cAAc,CAAC;AAC5B,cAAM,SAAS,KAAK,MAAM,IAAI,YAAY,IAAI,KAAK,SAAS;AAC5D,cAAM,SAAS,OAAO,MAAM;AAC5B,iBAAS,MAAM,MAAM;AACrB,cAAM,YAAY,eAAe,IAAI,IAAI,KAAK,CAAC;AAC/C,mBAAW,YAAY,WAAW;AAChC,kBAAQ,QAAQ,IAAI,OAAO,MAAM;AAAA,QACnC;AAAA,MACF;AAAA,IACF;AAEA,QAAI,QAAQ,SAAS,QAAW;AAC9B,aAAO,QAAQ,CAAC;AAAA,IAClB;AACA,WAAO;AAAA,EACT;AACF,CAAC;","names":[]}

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "qwen-embedder",
-  "version": "0.1.1",
-  "description": "Local text embedding with Qwen ONNX model via Transformers.js",
+  "version": "0.1.4",
+  "description": "A lightweight, optimized local text embedding generation using qwen model",
   "type": "module",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
@@ -29,12 +29,14 @@
     "nlp",
     "text-embedding",
     "local-ai",
-    "feature-extraction"
+    "feature-extraction",
+    "qwen-embedder"
   ],
   "license": "MIT",
   "dependencies": {
     "@huggingface/transformers": "^4.2.0",
-    "p-queue": "^9.3.0"
+    "p-queue": "^9.3.0",
+    "synckit": "^0.11.13"
   },
   "devDependencies": {
     "@types/node": "^25.9.3",

package/src/embedder.ts

@@ -3,6 +3,8 @@ import PQueue from 'p-queue'
 import { homedir } from 'os'
 import { join } from 'path'
 import type { EmbedderOptions } from './types.js'
+import { SyncEmbedder } from './sync-embedder.js'
+import type { SyncEmbedderOptions } from './sync-embedder.js'
 
 const MODEL_ID = 'onnx-community/Qwen3-Embedding-0.6B-ONNX'
 const CACHE_DIR = join(homedir(), '.qwenembedder', '.cache', 'models')
@@ -36,6 +38,10 @@ export class Embedder {
     this.queue = concurrency !== null ? new PQueue({ concurrency }) : null
   }
 
+  static createSync(options: SyncEmbedderOptions = {}): SyncEmbedder {
+    return new SyncEmbedder(options)
+  }
+
   static async create(options: EmbedderOptions = {}): Promise<Embedder> {
     const cacheSize = options.cacheSize ?? 100
     const concurrency = resolveConcurrency(options)
@@ -78,7 +84,8 @@ export class Embedder {
     }
 
     const results: (number[] | undefined)[] = new Array(texts.length)
-    const uncached: number[] = []
+    const uncachedByText = new Map<string, number[]>()
+    const uncachedTexts: string[] = []
 
     for (let i = 0; i < texts.length; i++) {
       const t = texts[i]
@@ -87,23 +94,33 @@ export class Embedder {
       }
       const cached = this.cache.get(t)
       if (cached !== undefined) {
-        results[i] = cached
+        results[i] = cached.slice()
       } else {
-        uncached.push(i)
+        const existing = uncachedByText.get(t)
+        if (existing !== undefined) {
+          existing.push(i)
+        } else {
+          uncachedByText.set(t, [i])
+          uncachedTexts.push(t)
+        }
       }
     }
 
-    if (uncached.length === 0) {
+    if (uncachedTexts.length === 0) {
       return results as number[][]
     }
 
-    const uncachedTexts = uncached.map(i => texts[i])
     const inferred = await this.runInference(uncachedTexts)
 
-    for (let j = 0; j < uncached.length; j++) {
+    for (let j = 0; j < uncachedTexts.length; j++) {
+      const text = uncachedTexts[j]
       const vector = inferred[j]
-      this.setCache(uncachedTexts[j], vector)
-      results[uncached[j]] = vector
+      const stored = vector.slice()
+      this.setCache(text, stored)
+      const positions = uncachedByText.get(text) ?? []
+      for (const position of positions) {
+        results[position] = stored.slice()
+      }
     }
 
     return results as number[][]
@@ -112,7 +129,7 @@ export class Embedder {
   private async runInference(text: string): Promise<number[]>
   private async runInference(texts: string[]): Promise<number[][]>
   private async runInference(input: string | string[]): Promise<number[] | number[][]> {
-    const exec = () => this.pipe(input, { pooling: 'mean', normalize: true })
+    const exec = () => this.pipe(input, { pooling: 'last_token', normalize: true })
 
     const output = this.queue
       ? await this.queue.add(exec)
@@ -135,12 +152,13 @@ export class Embedder {
   private setCache(key: string, value: number[]): void {
     if (this.maxCacheSize === 0) return
 
+    const stored = value.slice()
     if (this.cache.size >= this.maxCacheSize) {
       const oldest = this.cache.keys().next()
       if (!oldest.done) {
         this.cache.delete(oldest.value)
       }
     }
-    this.cache.set(key, value)
+    this.cache.set(key, stored)
   }
 }

package/src/index.ts

@@ -1,2 +1,4 @@
 export { Embedder } from './embedder.js'
+export { SyncEmbedder } from './sync-embedder.js'
 export type { EmbedderOptions } from './types.js'
+export type { SyncEmbedderOptions } from './sync-embedder.js'

package/src/sync-embedder.ts

@@ -0,0 +1,34 @@
+import { createSyncFn } from 'synckit'
+import { fileURLToPath } from 'url'
+import { dirname, resolve } from 'path'
+
+const __dirname = dirname(fileURLToPath(import.meta.url))
+const workerPath = resolve(__dirname, 'sync-worker.js')
+
+type SyncFn = {
+  (command: { _cmd: 'init'; cacheSize?: number }): void
+  (command: { _cmd: 'embed'; text: string }): number[]
+  (command: { _cmd: 'embed'; texts: string[] }): number[][]
+}
+
+export interface SyncEmbedderOptions {
+  cacheSize?: number
+}
+
+export class SyncEmbedder {
+  private fn: SyncFn
+
+  constructor(options: SyncEmbedderOptions = {}) {
+    this.fn = createSyncFn(workerPath) as SyncFn
+    this.fn({ _cmd: 'init', cacheSize: options.cacheSize ?? 100 })
+  }
+
+  embedSync(text: string): number[]
+  embedSync(texts: string[]): number[][]
+  embedSync(input: string | string[]): number[] | number[][] {
+    if (Array.isArray(input)) {
+      return this.fn({ _cmd: 'embed', texts: input })
+    }
+    return this.fn({ _cmd: 'embed', text: input })
+  }
+}

package/src/sync-worker.ts

@@ -0,0 +1,96 @@
+import { pipeline } from '@huggingface/transformers'
+import { runAsWorker } from 'synckit'
+import { homedir } from 'os'
+import { join } from 'path'
+
+type PipeFn = (texts: string[], options?: Record<string, unknown>) => Promise<{ data: Float32Array; dims: number[] }>
+
+const MODEL_ID = 'onnx-community/Qwen3-Embedding-0.6B-ONNX'
+const CACHE_DIR = join(homedir(), '.qwenembedder', '.cache', 'models')
+
+let pipe: PipeFn | null = null
+let cache: Map<string, number[]> | null = null
+let maxCacheSize = 100
+
+function setCache(key: string, value: number[]): void {
+  if (!cache) return
+  if (maxCacheSize === 0) return
+  if (cache.size >= maxCacheSize) {
+    const oldest = cache.keys().next()
+    if (!oldest.done) {
+      cache.delete(oldest.value)
+    }
+  }
+  cache.set(key, value)
+}
+
+type InitCommand = { _cmd: 'init'; cacheSize?: number }
+type EmbedCommand = { _cmd: 'embed'; text?: string; texts?: string[] }
+
+runAsWorker(async (command: InitCommand | EmbedCommand) => {
+  if (command._cmd === 'init') {
+    maxCacheSize = command.cacheSize ?? 100
+    cache = new Map()
+    return
+  }
+
+  if (!pipe) {
+    pipe = (await pipeline('feature-extraction', MODEL_ID, {
+      cache_dir: CACHE_DIR,
+      dtype: 'q8',
+    })) as unknown as PipeFn
+  }
+
+  if (command._cmd === 'embed') {
+    const texts = command.texts ?? (command.text ? [command.text] : [])
+
+    if (texts.length === 0) {
+      throw new TypeError('Input must be a non-empty string')
+    }
+
+    const results: (number[] | undefined)[] = new Array(texts.length)
+    const uncachedByText = new Map<string, number[]>()
+    const uncachedTexts: string[] = []
+
+    for (let i = 0; i < texts.length; i++) {
+      const t = texts[i]
+      if (typeof t !== 'string' || t.length === 0) {
+        throw new TypeError('Input must be a non-empty string')
+      }
+      const cached = cache?.get(t)
+      if (cached !== undefined) {
+        results[i] = cached.slice()
+      } else {
+        const existing = uncachedByText.get(t)
+        if (existing !== undefined) {
+          existing.push(i)
+        } else {
+          uncachedByText.set(t, [i])
+          uncachedTexts.push(t)
+        }
+      }
+    }
+
+    if (uncachedTexts.length > 0) {
+      const output = await pipe!(uncachedTexts, { pooling: 'last_token', normalize: true })
+      const data = Array.from(output.data) as number[]
+      const hiddenDim = output.dims[output.dims.length - 1]
+
+      for (let j = 0; j < uncachedTexts.length; j++) {
+        const text = uncachedTexts[j]
+        const vector = data.slice(j * hiddenDim, (j + 1) * hiddenDim)
+        const stored = vector.slice()
+        setCache(text, stored)
+        const positions = uncachedByText.get(text) ?? []
+        for (const position of positions) {
+          results[position] = stored.slice()
+        }
+      }
+    }
+
+    if (command.text !== undefined) {
+      return results[0] as number[]
+    }
+    return results as number[][]
+  }
+})