@nodd/node

Run a nodd model in Node with transformers.js on onnxruntime-node (native CPU). Same API as @nodd/browser; the export folder is read from disk.

npm install @nodd/node

import { nodd } from "@nodd/node";

const m = await nodd.load("./models/comment_moderation/v3"); // a `nodd export` folder
const d = await m.decide("Buy cheap followers at ...");
// { label: "spam", probabilities: {...}, confidence: 0.98, source: "micro", ... }
if (!m.isConfident(d)) { /* below the calibrated threshold: escalate */ }

Options: dtype: "q8" | "fp32" (fp32 only if the export includes onnx/model.onnx).

Why a separate package from the browser one: in Node there is no Web Worker, Cache API or same-origin rule, and onnxruntime-node runs the model natively (≈1 ms per input for a MiniLM-L6 q8 on an M2 Pro), so none of the browser plumbing or .wasm files are needed. Requires Node ≥ 20.