Hibbem's text compression uses a table of letter statistics (which bytes tend to follow
which) counted from example sentences of the Tatoeba project, in 21 languages, as
downloaded on 10 October 2026.

Tatoeba: https://tatoeba.org
The sentences are licensed under the Creative Commons Attribution 2.0 France licence
(CC BY 2.0 FR): https://creativecommons.org/licenses/by/2.0/fr/

No sentence is included in Hibbem; only the counts made from them. How the table was made:
tools/textmodel/build_model.py in Hibbem's source code.
