CircularBalls/plaincode-cnl-100k

Plaincode CNL 100k This dataset is generated with Plaincode by streaming Python source files, projecting each accepted Python module/function into the current controlled-natural-language surfaces, and enforcing strict exact Python roundtrip proof for every emitted language. The public row schema is one wide row per accepted Python source: source first, then English/Spanish/French/Mandarin/Hindi/Portuguese CNL columns, then selected metadata. Included surfaces… See the full description on the dataset page: https://huggingface.co/datasets/CircularBalls/plaincode-cnl-100k.