Add composite license notes and metadata source

LICENSE

@@ -0,0 +1,48 @@
+Talkie YaRN 32k Gutenberg Pre-1931 Dataset License Notes
+
+This repository is a generated dataset assembled from multiple upstream
+sources. It does not claim a new copyright over public-domain source text.
+
+1. Text rows
+
+The `text` fields in `train.jsonl` and `validation.jsonl` are selected from:
+
+  common-pile/project_gutenberg_filtered
+  https://huggingface.co/datasets/common-pile/project_gutenberg_filtered
+
+The upstream dataset card describes this corpus as English Project Gutenberg
+books marked as Public Domain according to the provided metadata, plus books
+from PG-19. This dataset further filtered rows to records whose `rights`
+metadata contains "Public Domain". The selected text is distributed here under
+the public-domain status reported by the upstream source metadata.
+
+Project Gutenberg and upstream metadata can contain mistakes or jurisdictional
+differences. Users should verify source rights for their own jurisdiction and
+use case.
+
+2. Publication-year metadata
+
+The file `metadata/gutenberg_publication_years.csv` comes from the Kaggle
+dataset:
+
+  Gutenberg Book Metadata with Publication Years
+  https://www.kaggle.com/datasets/yuvalschwartz/gutenberg-book-metadata-with-publication-years
+
+The Kaggle dataset page lists the license as "CC0: Public Domain". The metadata
+CSV is therefore included here under CC0-1.0 / public-domain dedication terms.
+
+3. Generated manifests, logs, and preparation script
+
+The summary JSON files and reconstruction log document how this snapshot was
+generated. The preparation script is included for reproducibility. It is
+provided under Apache License 2.0, consistent with the associated Talkie
+tooling.
+
+4. Composite license metadata
+
+Because the repository combines public-domain source texts, CC0 publication-year
+metadata, and an Apache-2.0 preparation script, the Hugging Face card uses:
+
+  license: other
+  license_name: public-domain-gutenberg-texts-and-cc0-metadata
+  license_link: LICENSE

README.md

@@ -2,6 +2,8 @@
 language:
 - en
 license: other
+license_name: public-domain-gutenberg-texts-and-cc0-metadata
+license_link: LICENSE
 task_categories:
 - text-generation
 pretty_name: Talkie YaRN 32k Gutenberg Pre-1931 265M
@@ -28,11 +30,13 @@ This dataset is the continued-pretraining corpus used for the Talkie YaRN 32k
 context-extension experiments, including the recommended checkpoint
 [`xlr8harder/talkie-1930-13b-yarn-32k-tf`](https://huggingface.co/xlr8harder/talkie-1930-13b-yarn-32k-tf).
 
-It was generated from `common-pile/project_gutenberg_filtered`, joined with a
-local Gutenberg publication-year metadata table, then filtered to English
-public-domain books with publication years from 1500 through 1930. Project
-Gutenberg boilerplate markers were stripped and each book was capped at 500,000
-characters.
+It was generated from
+[`common-pile/project_gutenberg_filtered`](https://huggingface.co/datasets/common-pile/project_gutenberg_filtered),
+joined with a Gutenberg publication-year metadata table from Kaggle,
+[`yuvalschwartz/gutenberg-book-metadata-with-publication-years`](https://www.kaggle.com/datasets/yuvalschwartz/gutenberg-book-metadata-with-publication-years),
+then filtered to English public-domain books with publication years from 1500
+through 1930. Project Gutenberg boilerplate markers were stripped and each book
+was capped at 500,000 characters.
 
 ## Splits
 
@@ -114,7 +118,11 @@ python scripts/prepare_gutenberg_with_years.py \
 
 ## License Notes
 
-The source rows selected here are marked `Public Domain` in
-`common-pile/project_gutenberg_filtered`. This dataset card uses `other` because
-the corpus is a generated selection from upstream sources plus a metadata join;
-verify source rights and downstream requirements for your use case.
+This repository uses `license: other` with a custom license description because
+it combines multiple upstream sources. The text rows inherit the public-domain
+status reported by `common-pile/project_gutenberg_filtered` for the selected
+Project Gutenberg rows. The included publication-year metadata CSV comes from
+the Kaggle dataset
+[`yuvalschwartz/gutenberg-book-metadata-with-publication-years`](https://www.kaggle.com/datasets/yuvalschwartz/gutenberg-book-metadata-with-publication-years),
+which is listed as CC0: Public Domain. See [`LICENSE`](./LICENSE) for the
+component-level license notes.