• Two tasks: one text/NLP, one audio processing/speech-to-text.
• Everything in build/ is generated or an upstream library.
• Everything in util/ is tooling for building and preparing the benchmark.

Licenses

• Transformers.js: Apache 2.0, https://github.com/huggingface/transformers.js/blob/main/LICENSE
• ONNX runtime: MIT, https://github.com/microsoft/onnxruntime/blob/main/LICENSE
• text-encoding Polyfill: Unlicense OR Apache 2.0, https://github.com/inexorabletash/text-encoding/blob/master/LICENSE.md
• Model DistilBERT base uncased finetuned SST-2: Apache 2.0, https://huggingface.co/distilbert/distilbert-base-uncased-finetuned-sst-2-english
• Model openai/whisper-tiny.en: Apache 2.0, https://huggingface.co/openai/whisper-tiny.en
• Audio file for speech-to-text task: Public domain, https://www.jfklibrary.org/learn/about-jfk/historic-speeches/inaugural-address