Specialty multilingual training data for
enterprise AI applications.
Chedai produces structured multilingual translation datasets with token-level metadata for enterprise AI training applications.
Ancient Greek and Latin into English are in production. The pipeline is source- and target-language agnostic by design; further ancient sources and modern target languages are in development.
Every text is rendered in two registers under a single editorial standard. Every token carries 37 fields of linguistic metadata. Every source text is public domain; its edition, provenance, and license status are documented per text.
| Ancient Greek | Latin | |
|---|---|---|
| Form | μῆνιν | arma |
| Lemma | μῆνις | arma |
| Part of speech | noun | noun |
| Morphology | acc · sg · fem | acc · pl · neut |
| Gloss | wrath, rage | arms, weapons |