Corpus Kirkensis: A Permissively-Licensed Text Corpus for LLM Provenance

The Corpus Kirkensis is a text dataset primarily composed of Libre content, with a few non-permissive texts covered by copyright exceptions. It was created by author rtc2022 and last updated on Hugging Face on 2026-05-30. The dataset is intended to address data provenance and output generation issues in the large language model space.

Use Cases

• Training language models based on permissively-licensed text content.
• Studying data provenance and licensing in LLM training datasets.
• Generating creative or interesting text output based on a curated corpus of Libre content.

Strengths

• Primarily composed of Libre content, suggesting a focus on permissive licensing.
• Explicitly created to address data provenance, a key concern in LLM development.
• Last updated on 2026-05-30, indicating recent maintenance.

Limitations

• The exact size, row count, and file formats are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• The description is incomplete, referencing a full description on an external page.

Provenance

Source

huggingface, author rtc2022

Collection Method

Curated collection of Libre content.

Freshness

Last updated 2026-05-30 16:23:47.