Licences
Code is Apache 2.0. The tokenizer keeps the terms of the tokenizer it extends, and data stays with its sources. Each notice is below.
Datasets We Use
FLORES+, NTREX-128 and Belebele under CC BY-SA 4.0; FineWeb-2 under ODC-By 1.0; WAXAL under CC BY 4.0 and CC BY-SA 4.0; MAFAND-MT (Masakhane). Downloaded, never redistributed.
Fonts
We use Geist and Geist Mono, free for personal and commercial use under the SIL Open Font License 1.1. Both are Google Fonts, and load with the template's own styles. You can check the licence for each font using the links:
Code and Tokenizer
Tokenizer code is licensed under the Apache License 2.0. The tokenizer is derived from the Gemma 3 tokenizer: Gemma is provided under and subject to the Gemma Terms of Use at ai.google.dev/gemma/terms. The Llama 3.1 variant is under the Llama 3.1 Community License.
Logo
The cowrie mark and the drawn nzinga wordmark are Nzinga's own, made by HQ Protocol. The name honours Njinga Mbande; the cowrie stands for payment and speech. Ask before using either.
Template licence
This site reskins a licensed website template, used under its Marketplace licence.