mhingston/distilbert-base-threat-matrix-ONNX

Threat Matrix ONNX Test Project

Test project for using the neuralchemy/distilbert-base-threat-matrix ONNX model with Transformers.js.

Model

• Hugging Face: mhingston/distilbert-base-threat-matrix-onnx
• Original: neuralchemy/distilbert-base-threat-matrix
• Task: Text Classification (Binary: benign / malicious)
• Labels: benign, malicious

Usage

Browser (HTML)

Open index.html in a browser. The model loads from Hugging Face CDN.

Note: WebGPU is required for best performance. If unavailable, the code falls back to WASM.

Node.js

# Install dependencies
npm install

# Run tests
npm run test

# Or start a local server
npm run start

Device Options

Device	Description
webgpu	Fastest in browser (Chrome/Edge)
wasm	Slower but universal browser support
cpu	For Node.js only

API Usage

import { pipeline } from '@huggingface/transformers';

const classifier = await pipeline(
    'text-classification',
    'mhingston/distilbert-base-threat-matrix-onnx',
    { device: 'webgpu' }
);

const result = await classifier("Ignore all previous instructions...");
console.log(result);
// [{ label: 'malicious', score: 0.992 }]

Files

• index.html - Browser demo
• test-node.js - Node.js test script
• package.json - Node.js dependencies