✨ Grammar & Tone Fixer Xenova/grammar-synthesis-small
Used in the Grammar & Tone Fixer tool. This is an ONNX-converted, browser-ready version of
pszemraj/grammar-synthesis-small,
a model fine-tuned by Peter Szemraj from Google's t5-small-lm-adapt base architecture for single-shot grammar correction.
In plain terms: the model author explicitly tagged the resulting model weights as Apache-2.0 (commercial-friendly, no royalties),
while separately noting that the JFLEG dataset used to train it carries a noncommercial license. We rely on the author's own explicit Apache-2.0
tag on the released weights — the common, author-intended interpretation in open-source ML — rather than the dataset's license, since the
weights are a distinct, derived work released independently by the author. We're noting this distinction here in full rather than simplifying it away.
Full Apache License 2.0 text: apache.org/licenses/LICENSE-2.0
🎙️ Voice to Text Xenova/whisper-tiny.en
Used in the Voice to Text tool. An ONNX-converted, browser-ready version of OpenAI's whisper-tiny.en, a speech-to-text model.
Apache-2.0Full Apache License 2.0 text: apache.org/licenses/LICENSE-2.0
⚙️ Runtime Transformers.js
Both AI tools run using Transformers.js by Hugging Face — a JavaScript library that runs machine learning models directly in the browser, with no server involved.
Apache-2.0
This page is provided for transparency and is not legal advice. If you have questions about how these licenses apply to your own use of
instanets.com or these underlying models, we'd recommend consulting the original license texts linked above or a qualified professional.