node package manager
Easy collaboration. Discover, share, and reuse code in your team. Create a free org »




A webpack loader for binary numpy .npy files


Configure the loader in your webpack.config.js under module > rules, e.g. like this:

  test: /\.(npy|npc)$/,
  exclude: /node_modules/,
  loader: 'numpy-loader',
  options: {
    outputPath: 'assets/data'

You can specify an outputPath that will be used during compilation.

You can now load .npy files as ndarrays in your JS code simply by requireing them. There are two modes, for small and large npy files respectively:

Small files

Load small .npy files directly from the webpack bundle using embed=true. These files will be base64 encoded and become a part of your webpack bundle.

let npyarray = require("numpy-loader?embed=true!./data/array_uint8.npy");
console.log("Loaded array in JS directly from packed module: " +;

Large files

You might not want to embed particularly large files in your webpack bundle. You can specify embed=false (or nothing; it's the default) to copy your .npy files to your output directory and load them as binary files from your server at runtime instead. Thus you need to use the #load() callback like this:

const npyarray = require("numpy-loader?embed=false!./data/array_uint8.npy")
npyarray.load( (array) => {
  console.log("Loaded an .npy array in JS!");