Traditional ASCII art subset from https://huggingface.co/datasets/apehex/ascii-art at revision 0669726cf305d48dd0e02426d136c9e92a2f476e. Only asciiart/train is used. Images and generated graffiti are excluded. Dataset card license: AGPL-3.0. Artist signatures in the text are retained. Source file links and SHA-256 checksums are in source-files.json; normalization and exclusion code is supplied as importer.py. Original source documented by the dataset is https://www.asciiart.eu/. Changes: preserve line structure, CRLF to LF, 8-column tab expansion, strip trailing whitespace, normalize for duplicate comparison, exclude adult-tagged entries.
