Add LangChain.js adapter for OpenVINO (#973)

* Add LangChain.js adapter for OpenVINO

Signed-off-by: Kirill Suvorov <[email protected]>

* Move into modules directory

* Update dependency

* npm pkg fix

Signed-off-by: Kirill Suvorov <[email protected]>

* Remove outdated docs

* Update licenses

---------

Signed-off-by: Kirill Suvorov <[email protected]>

Author: Retribution98

README.md

@@ -38,6 +38,8 @@ Additional build instructions are available for the following modules:
 * [**nvidia_plugin**](./modules/nvidia_plugin/README.md)
 * [**custom_operations**](./modules/custom_operations/README.md)
 * [**ollama_OpenVINO**](./modules/ollama_openvino)
+* [**openvino-langchain**](./modules/openvino-langchain): LangChain.js integrations for OpenVINO™
+
 ## Update the repository documentation
 In order to keep a clean overview containing all contributed modules, the following files need to be created/adapted:
 

modules/openvino-langchain/.gitignore

@@ -0,0 +1,4 @@
+dist
+node_modules
+sample/models
+types

modules/openvino-langchain/README.md

@@ -0,0 +1,101 @@
+# OpenVINO™ LangChain.js adapter
+
+This package contains the LangChain.js integrations for OpenVINO™
+
+> **Disclaimer**
+> It's preview version, do not use it on production!
+
+## Introduction
+
+OpenVINO is an open-source toolkit for deploying performant AI solutions. Convert, optimize, and run inference on local hardware utilizing the full potential of Intel® hardware.
+
+## Installation and Setup
+
+See [this section](https://js.langchain.com/docs/how_to/installation#installing-integration-packages) for general instructions on installing integration packages.
+
+```bash
+npm install openvino-langchain
+```
+
+### Export your model to the OpenVINO™ IR
+
+In order to use OpenVINO, you need to convert and compress the text generation model into the [OpenVINO IR format](https://docs.openvino.ai/2025/documentation/openvino-ir-format.html).
+
+Tested compatibility with:
+  - BAAI/bge-small-en-v1.5 (Embeddings model)
+  - openlm-research/open_llama_7b_v2
+  - meta-llama/Llama-2-13b-chat-hf
+  - microsoft/Phi-3.5-mini-instruct
+
+#### Use HuggingFace Hub
+
+Pre-converted and pre-optimized models are available under the [LLM collections](https://huggingface.co/collections/OpenVINO/llm-6687aaa2abca3bbcec71a9bd) in the [OpenVINO Toolkit](https://huggingface.co/OpenVINO) organization.
+
+To export another [model](https://huggingface.co/docs/optimum/main/en/intel/openvino/models) hosted on the [HuggingFace Hub](https://huggingface.co/models) you can use [OpenVINO space](https://huggingface.co/spaces/echarlaix/openvino-export). After conversion, a repository will be pushed under your namespace, this repository can be either public or private.
+
+#### Use the Optimum Intel
+
+[Optimum Intel](https://github.com/huggingface/optimum-intel) provides a simple interface to optimize your Transformers and Diffusers models and convert them to the OpenVINO Intermediate Representation (IR) format.
+
+Firstly install Optimum Intel for OpenVINO:
+
+```bash
+pip install --upgrade --upgrade-strategy eager "optimum[openvino]"
+```
+
+Then you download and convert a model to OpenVINO:
+
+```bash
+optimum-cli export openvino --model <model_id> --trust-remote-code <exported_model_name>
+```
+> **Note:** Any model_id, for example "TinyLlama/TinyLlama-1.1B-Chat-v1.0", or the path to a local model file can be used.
+
+Optimum-Intel API also provides out-of-the-box model optimization through weight compression using NNCF which substantially reduces the model footprint and inference latency:
+
+```bash
+optimum-cli export openvino --model "TinyLlama/TinyLlama-1.1B-Chat-v1.0" --weight-format int4 --trust-remote-code "TinyLlama-1.1B-Chat-v1.0"
+```
+
+## LLM
+
+This package contains the `GenAI` class, which is the recommended way to interact with models optimized for the OpenVINO toolkit.
+
+**GenAI Parameters**
+
+| Name  | Type | Required | Description |
+| ----- | ---- |--------- | ----------- |
+| modelPath | string | ✅ | Path to the directory containing model xml/bin files and tokenizer |
+| device | string | ❌ | Device to run the model on (e.g., CPU, GPU). |
+| generationConfig | [GenerationConfig](https://github.com/openvinotoolkit/openvino.genai/blob/master/src/js/lib/utils.ts#L107-L110) | ❌ | Structure to keep generation config parameters. |
+
+```typescript
+import { GenAI } from "openvino-langchain";
+
+const model = new GenAI({
+    modelPath: "path-to-model",
+    device: "CPU",
+    generationConfig: {
+        "max_new_tokens": 100,
+    },
+  });
+const response = await model.invoke("Hello, world!");
+```
+
+## Text Embedding Model
+
+This package also adds support for OpenVINO's embeddings model.
+
+| Name  | Type | Required | Description |
+| ----- | ---- |--------- | ----------- |
+| modelPath | string | ✅ | Path to the directory containing embeddings model |
+| device | string | ❌ | Device to run the embeddings model on (e.g., CPU, GPU). |
+
+```typescript
+import { OvEmbeddings } from "openvino-langchain";
+
+const embeddings = new OvEmbeddings({
+    modelPath: "path-to-model",
+    device: "CPU",
+});
+const res = await embeddings.embedQuery("Hello world");
+```

modules/openvino-langchain/eslint.config.ts

@@ -0,0 +1,45 @@
+import globals from 'globals';
+import pluginJs from '@eslint/js';
+import { parser, ConfigArray } from 'typescript-eslint';
+
+export default [
+  pluginJs.configs.recommended,
+  {
+    ignores: [ 'types/', 'dist/' ],
+  },
+  {
+    files: ['**/*.{js,mjs,cjs,ts}'],
+    languageOptions: {
+      globals: globals.node,
+      parser: parser,
+    },
+    rules: {
+      'semi': ['error'],
+      'no-var': ['error'],
+      'max-len': ['error', { 'ignoreUrls': true }],
+      'eol-last': ['error'],
+      'indent': ['error', 2],
+      'camelcase': ['error'],
+      'semi-spacing': ['error'],
+      'arrow-spacing': ['error'],
+      'comma-spacing': ['error'],
+      'no-multi-spaces': ['error'],
+      'quotes': ['error', 'single'],
+      'no-trailing-spaces': ['error'],
+      'space-before-blocks': ['error'],
+      'newline-before-return': ['error'],
+      'comma-dangle': ['error', 'always-multiline'],
+      'space-before-function-paren': ['error', {
+        named: 'never',
+        anonymous: 'never',
+        asyncArrow: 'always',
+      }],
+      'key-spacing': ['error', { beforeColon: false }],
+      'no-multiple-empty-lines': ['error', { max: 1, maxBOF: 0, maxEOF: 0 }],
+      'keyword-spacing': ['error', { overrides: { catch: { after: false } } }],
+      'prefer-destructuring': ['error', { object: true, array: false }],
+      'no-explicit-any': 0,
+      'no-unused-vars': ['error', { args: 'none' } ],
+    },
+  },
+] satisfies ConfigArray;

modules/openvino-langchain/jest.config.cjs

@@ -0,0 +1,22 @@
+/** @type {import('ts-jest').JestConfigWithTsJest} */
+
+module.exports = {
+  preset: 'ts-jest/presets/default-esm',
+  testEnvironment: './jest.env.cjs',
+  modulePathIgnorePatterns: ['dist/', 'docs/'],
+  moduleNameMapper: {
+    '^(\\.{1,2}/.*)\\.(c)*js$': '$1',
+  },
+  transform: {
+    '^.+\\.tsx?$': ['@swc/jest'],
+  },
+  transformIgnorePatterns: [
+    '/node_modules/',
+    '\\.pnp\\.[^\\/]+$',
+    './scripts/jest-setup-after-env.js',
+  ],
+  setupFiles: ['dotenv/config'],
+  setupFilesAfterEnv: ['./scripts/jest-setup-after-env.js'],
+  testTimeout: 20_000,
+  collectCoverageFrom: ['src/**/*.ts'],
+};

modules/openvino-langchain/jest.env.cjs

@@ -0,0 +1,12 @@
+const { TestEnvironment } = require('jest-environment-node');
+
+class AdjustedTestEnvironmentToSupportFloat32Array extends TestEnvironment {
+  constructor(config, context) {
+    // Make `instanceof Float32Array` return true in tests
+    // to avoid https://github.com/xenova/transformers.js/issues/57 and https://github.com/jestjs/jest/issues/2549
+    super(config, context);
+    this.global.Float32Array = Float32Array;
+  }
+}
+
+module.exports = AdjustedTestEnvironmentToSupportFloat32Array;