NER Benchmark

Benchmarks Firefox-compatible ONNX NER models for speed and person-name recall, with a spaCy en_core_web_lg baseline for comparison.

Setup

ONNX benchmark (Node.js):

npm install

spaCy baseline (Python):

python3 -m venv .venv
.venv/bin/pip install spacy
.venv/bin/python3 -m spacy download en_core_web_lg

Run

ONNX benchmark:

node benchmark.mjs

Models are downloaded from HuggingFace on first run and cached locally. Expect a few minutes the first time; subsequent runs are fast.

spaCy baseline:

.venv/bin/python3 spacy_benchmark.py

Results

See context/2026-03-ner.md for the full analysis.