|
|
| """Generate Dynaword documentation: per-source datasheets + README + CHANGELOG + LICENSE.
|
|
|
| Reads sources.py + data/<source>/<source>.stats.json (written by build_dynaword.py).
|
| Implements the "Documented" principle (datasheets, Gebru et al. 2021) and the
|
| aggregate README table (paper 2508.02271).
|
| """
|
| from __future__ import annotations
|
| import json, sys
|
| from pathlib import Path
|
|
|
| sys.path.insert(0, str(Path(__file__).resolve().parent))
|
| from sources import SOURCES, EXCLUDED, ADDED
|
|
|
| ROOT = Path(__file__).resolve().parent.parent
|
| VERSION = "0.2.4"
|
| RELEASE_DATE = "2026-08-07"
|
| CONTACT = "k.wikiel@gmail.com"
|
|
|
|
|
| DISCLAIMER = f"""## Personal & sensitive data
|
| This corpus contains **only** text that its upstream sources already published
|
| under open licenses or as official public-domain record. It therefore includes
|
| names and statements of **public figures acting in a public capacity** — e.g.
|
| parliamentary speakers (PPC), authorities named in legal acts (EUR-Lex), and
|
| people described in encyclopedic articles (Wikipedia/Wikisource). No private,
|
| non-public personal data was collected or added. If you are a data subject and
|
| want content concerning you removed, contact **{CONTACT}** — it will be dropped
|
| from the next version (see retroactive-removal policy below).
|
|
|
| ## Disclaimer & legal
|
| - **Provenance in good faith.** Per-source licenses are reproduced *as documented
|
| by the upstream sources and by SpeakLeash* (the intermediate aggregator), to the
|
| best of our knowledge. We make no independent legal warranty about the copyright
|
| status of any individual document.
|
| - **No ownership claim.** This release is a *curated, license-reviewed, documented
|
| aggregation*. We claim no ownership of the underlying texts; rights remain with
|
| the original authors/rightsholders under their respective licenses.
|
| - **Provided "as is"**, without warranty of any kind, express or implied. This is
|
| not legal advice.
|
| - **Your compliance is yours.** Downstream users must satisfy each upstream
|
| license themselves — in particular **CC-BY-SA-4.0 attribution and share-alike**
|
| for derivatives of this dataset, and attribution to the upstream sources and to
|
| SpeakLeash.
|
| - **Notice-and-takedown.** Any source or rightsholder raising a substantiated
|
| objection can have material removed: contact **{CONTACT}**; it is dropped from
|
| the next version and recorded in the CHANGELOG. Removal is retroactive
|
| going-forward (prior immutable snapshots/commits may persist).
|
| """
|
|
|
|
|
| def load_stats(name):
|
| f = ROOT / "data" / name / f"{name}.stats.json"
|
| return json.loads(f.read_text()) if f.exists() else None
|
|
|
|
|
| def datasheet(name, cfg, st):
|
| license_rows = ""
|
| licenses = st.get("licenses") or {}
|
| if licenses:
|
| top = sorted(licenses.items(), key=lambda item: -item[1])[:20]
|
| license_rows = "\n\n## Per-document license metadata\n| license | documents |\n|---|---:|\n"
|
| license_rows += "\n".join(f"| `{k or 'UNKNOWN'}` | {v:,} |" for k, v in top)
|
| author_note = ""
|
| if "authors_with_value" in st:
|
| author_note = (
|
| f"\n\nAuthor metadata present for **{st.get('authors_with_value', 0):,}** "
|
| "documents. Empty values mean the upstream record did not expose a "
|
| "machine-readable author field."
|
| )
|
| legal_note = cfg.get("legal_note", "")
|
| if legal_note:
|
| legal_note = f"\n\n## Legal scope note\n{legal_note}"
|
| source_note = ""
|
| if st.get("stats_recomputed_from_parquet"):
|
| source_note = "\n\nStatistics were recomputed directly from the released parquet file."
|
| return f"""# {name}
|
|
|
| {cfg['pretty']}
|
|
|
| ## Dataset description
|
| - **Source (upstream):** {cfg['upstream']}
|
| - **Domain:** {cfg['domain']}
|
| - **Language:** Polish (pl)
|
| - **License:** `{cfg['license']}`
|
| - **Created (range):** {cfg['created']}
|
| - **Added:** {ADDED}
|
|
|
| ## Licensing — traceable basis
|
| {cfg['traceable']}
|
|
|
| ## Provenance
|
| {cfg.get('provenance', f"Pulled from SpeakLeash's public redistribution "
|
| f"(`speakleash-ds-pub`, key `{cfg.get('speakleash_key')}`) of the upstream source "
|
| f"above. SpeakLeash credited as intermediate aggregator; upstream "
|
| f"license/attribution preserved.")}
|
|
|
| ## Statistics
|
| | documents | characters | tokens (tiktoken proxy) |
|
| |---:|---:|---:|
|
| | {st['kept']:,} | {st['chars']:,} | {st['tokens']:,} |
|
| {license_rows}{author_note}{legal_note}{source_note}
|
|
|
| ## Filters applied (build_dynaword.py)
|
| Minimal, per Dynaword guidelines (heavy filtering left to downstream use):
|
| - drop documents < 200 chars: **{st.get('drop_short', 0):,}**
|
| - drop non-Polish (diacritic ratio): **{st.get('drop_lang', 0):,}**
|
| - exact cross-source dedup (sha1): **{st.get('drop_dup', 0):,}**
|
| - OCR alpha-ratio < 0.70 (OCR sources only): **{st.get('drop_ocr', 0):,}**
|
| - read {st.get('read', st['kept']):,} → kept {st['kept']:,}
|
|
|
| Token counts are a fast tiktoken (cl100k) proxy (~1% off Llama-3); the canonical
|
| Llama-3 count is computed at release.
|
| """
|
|
|
|
|
| def main():
|
| rows, tot_doc, tot_tok, tot_chr = [], 0, 0, 0
|
| for name, cfg in SOURCES.items():
|
| st = load_stats(name)
|
| if not st:
|
| print(f" ! no stats for {name}"); continue
|
|
|
|
|
| if not cfg.get("custom_datasheet"):
|
| (ROOT / "data" / name / f"{name}.md").write_text(datasheet(name, cfg, st))
|
| rows.append((name, cfg, st))
|
| tot_doc += st["kept"]; tot_tok += st["tokens"]; tot_chr += st["chars"]
|
| rows.sort(key=lambda r: -r[2]["tokens"])
|
|
|
| tbl = "\n".join(
|
| f"| [{n}](data/{n}/{n}.md) | {c['pretty']} | `{c['license']}` | "
|
| f"{s['kept']:,} | {s['tokens']/1e6:,.1f}M |"
|
| for n, c, s in rows)
|
| excl = "\n".join(f"| `{k}` | {v} |" for k, v in EXCLUDED.items())
|
| phrase_frequency = ""
|
| phrase_path = ROOT / "artifacts" / "pattern_frequency_hf_snippet.md"
|
| if phrase_path.exists():
|
| snippet = phrase_path.read_text().replace(
|
| "## Phrase frequency in corpus (token-normalized)\n\n", ""
|
| )
|
| phrase_frequency = (
|
| "\n## Results\n\n"
|
| "### Corpus phrase frequency (normalized by tokens)\n\n"
|
| "Raw counts and token-normalized shares are regenerated from the "
|
| "current parquet files with `src/pattern_frequency_report.py`.\n\n"
|
| f"{snippet}"
|
| )
|
|
|
| readme = f"""---
|
| license: cc-by-sa-4.0
|
| language:
|
| - pl
|
| pretty_name: Polish DynaWord
|
| task_categories:
|
| - text-generation
|
| size_categories:
|
| - 1M<n<10M
|
| tags:
|
| - polish
|
| - pretraining
|
| - dynaword
|
| ---
|
|
|
| # Polish DynaWord
|
|
|
| A continuously developed, **openly-licensed**, human-text Polish corpus — a Polish
|
| edition in the [Dynaword](https://huggingface.co/datasets/danish-foundation-models/danish-dynaword)
|
| family (Enevoldsen et al., [arXiv:2508.02271](https://arxiv.org/abs/2508.02271)).
|
|
|
| > **v{VERSION} stable** · {tot_doc:,} documents · **{tot_tok/1e9:.2f}B tokens**
|
| > (tiktoken proxy; canonical Llama-3 count at release) · {len(rows)} sources
|
| > Updated: **{RELEASE_DATE}**
|
|
|
| > **v0.3-dev experimental track** · quality/diversity workflow, source-gate
|
| > validation and candidate-data audits. This is development work, not a released
|
| > corpus version, and it does not replace the v{VERSION} stable parquets.
|
|
|
| > **Europeana validation artifact (2026-07-10)** · a separate reproducible
|
| > 810-document direct-ingestion sample stored under the legacy path
|
| > `previews/v0.3.1/europeana.parquet`. It is not a full dataset release.
|
|
|
| ## Releases
|
|
|
| ### Stable releases
|
|
|
| | version | status | documents | tokens | notes |
|
| |---|---|---:|---:|---|
|
| | `v0.2.4` | stable release | {tot_doc:,} | {tot_tok/1e9:.2f}B | Expanded `european_hplt_v3_pl` to WDS bins 10-5 (+1.49M docs / 2.04B tok); phone PII-scrub hardening (independent dual-lens) + cross-source dedup (zero-overlap). |
|
| | `v0.2.3` | previous stable | 2,710,974 | 6.88B | Adds community-contributed `european_hplt_v3_pl`, `global_voices` and `nkjp1m`. |
|
| | `v0.2.2` | previous stable | 2,579,963 | 6.36B | Added community-contributed `govpl`: 88,190 docs / 80.7M tokens. |
|
| | `v0.2.1` | previous stable | 2,491,773 | 6.28B | 12-source corpus with `license` and `author` metadata columns; added `1000_novels`. |
|
| | `v0.2.0` | previous stable | 2,490,773 | 6.22B | Provenance-first corpus from 11 open/official sources. |
|
|
|
| 
|
|
|
| ### Development and validation artifacts
|
|
|
| | name | type | scope | notes |
|
| |---|---|---|---|
|
| | `v0.3-dev` | experimental development track | workflow and candidate audits | Quality remix, legal-source caps, source QA, deduplication and direct-upstream ingestion. Not a corpus release. |
|
| | `Europeana validation artifact 2026-07-10` | reproducible sample | 810 documents / 140,042 tokens | Stored at the legacy path `previews/v0.3.1/europeana.parquet`; preserves per-record rights and creator metadata. Not a corpus release. |
|
|
|
| ### Version details
|
|
|
| #### v0.2.4 — current stable
|
|
|
| Expanded `european_hplt_v3_pl` from WDS bins 10+9 to **WDS bins 10-5** (added
|
| quality bins 8/7/6/5 and further WDS-9 shards): **+1,493,384 documents /
|
| 2,043,141,874 cl100k-proxy tokens**, 100% CC0-1.0. Includes a phone PII-scrub
|
| hardening pass (v12b) verified by two independent lenses (a systematic
|
| recall-gap masked by pipeline self-report was caught pre-release), and
|
| cross-source deduplication against the existing WDS-10+9 shards (exact-dup 0,
|
| near-dup overlap 0.0000%, byte-verified) — resolving the "downstream near-dedup
|
| remains" caveat from the original release.
|
|
|
| #### v0.2.3 — previous stable
|
|
|
| Community expansion release. Adds `european_hplt_v3_pl` from
|
| [PR #5](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/5)
|
| and `global_voices` from
|
| [PR #7](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/7),
|
| plus `nkjp1m` from
|
| [PR #8](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/8).
|
| The complete 16-source release contains **2,710,974 documents /
|
| 6,881,277,362 cl100k-proxy tokens**, summed from the released per-source parquet
|
| statistics.
|
|
|
| #### v0.2.2 — previous stable
|
|
|
| Added `govpl` from
|
| [PR #9](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/9):
|
| 88,190 Polish government press releases collected directly from 133 gov.pl
|
| ministry and agency subsites.
|
|
|
| #### v0.2.1 — previous stable
|
|
|
| Introduced the canonical eight-column release schema:
|
| `id, text, source, added, created, token_count, license, author`. Added
|
| `1000_novels` from
|
| [PR #1](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/1)
|
| and recomputed all release statistics from parquet files.
|
|
|
| #### v0.2.0 — initial stable corpus
|
|
|
| The provenance-first baseline: 11 reviewed open or official sources,
|
| 2,490,773 documents and 6.22B proxy tokens.
|
|
|
| #### v0.3-dev — experimental quality workflow
|
|
|
| Candidate-only work on a better-balanced training mixture: legal-source caps,
|
| temperature sampling, source QA, deduplication and direct-upstream ingestion.
|
| It is not a replacement for the stable corpus.
|
|
|
| #### Europeana validation artifact — 2026-07-10
|
|
|
| A separate reproducible 810-document Europeana sample preserving per-record
|
| rights and creator metadata. It does not change stable-release totals.
|
| `v0.3.1-preview` remains only as the legacy storage-path label.
|
|
|
| ## What this dataset contributes
|
| The raw texts come from existing open corpora (redistributed via SpeakLeash and,
|
| where applicable, fetched from upstream). **The value added here is the curation,
|
| not the bytes**, following the Dynaword methodology:
|
|
|
| 1. **License review per source** — each source vetted for an *openly-licensed,
|
| traceable* legal basis (documented in its datasheet); sources that fail the
|
| review are **excluded with a stated reason** (see table below), not silently
|
| kept. This is the core editorial work.
|
| 2. **Filtering & normalization** — minimal, reproducible gates (short-doc,
|
| non-Polish, exact cross-source dedup, OCR garble) applied uniformly to one
|
| clean schema: `id, text, source, added, created, token_count, license, author`.
|
| 3. **Documentation** — a datasheet per source (Gebru et al. 2021) + this card,
|
| so provenance and licensing are auditable rather than assumed.
|
| 4. **Reproducibility & versioning** — `src/` rebuilds the corpus from sources;
|
| new sources and removals are tracked in the CHANGELOG.
|
|
|
| Credit for the underlying texts belongs to the upstream sources and to SpeakLeash
|
| as the redistributing aggregator; this release does not claim ownership of them
|
| (see Disclaimer).
|
|
|
| ## Contributors
|
|
|
| | contributor | contribution | release / PR |
|
| |---|---|---|
|
| | [Kacper Wikieł](https://huggingface.co/kacperwikiel) | Project maintainer; corpus curation, source and license review, release engineering, documentation, validation and reproducible build workflow. | all releases |
|
| | [Bart Kobyliński](https://huggingface.co/bartoszkobylinski1) | Added `1000_novels`; expanded Biblioteka Nauki and Europeana ingestion with per-document license and author metadata. | `v0.2.1`, [PR #1](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/1) |
|
| | [Paweł Puzio](https://huggingface.co/ppuzio) | Built `govpl`: subsite discovery, direct ingestion pipeline, dataset artifact, contract tests and documentation. | `v0.2.2`, [PR #9](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/9) |
|
| | [Arkadiusz Słota](https://huggingface.co/Maggio33) | Built & expanded `european_hplt_v3_pl`: HPLT v3 WDS 10+9 (PR #5) then WDS 10-5 expansion (+1,493,384 docs / 2.04B tok), modular cleaning pipeline, phone-recall PII-scrub (v12b, independent dual-lens), cross-source dedup-vs-base (zero-overlap), validation and documentation. | `v0.2.3` (PR #5), `v0.2.4` |
|
| | [Dawid Majewski](https://huggingface.co/dawidmajewski) | Added the reviewed Global Voices Polish corpus with per-document author attribution and documentation. | `v0.2.3`, [PR #7](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/7) |
|
| | [1am](https://huggingface.co/1am) | Added `nkjp1m`: the manually annotated 1-million-word NKJP subcorpus, direct fetch/build pipeline, source documentation and release artifact. | `v0.2.3`, [PR #8](https://huggingface.co/datasets/SlayerLab/polish-dynaword/discussions/8) |
|
|
|
| Contributions are credited when they add a verifiable dataset artifact, pipeline,
|
| validation, documentation or release work. A merged discussion with no resulting
|
| files is not listed as a data contribution.
|
|
|
| ## Guiding principles
|
| 1. **Open & traceable licensing** — every source is *openly licensed* with a documented
|
| legal basis (see each datasheet's "traceable basis"), not a vague "public domain".
|
| 2. **Reproducibility** — `src/build_dynaword.py` rebuilds the corpus from sources.
|
| 3. **Documented** — a datasheet per source under `data/<source>/`.
|
| 4. **Extensibility** — versioned; new sources via PR.
|
|
|
| ## Sources
|
| | source | description | license | documents | tokens |
|
| |---|---|---|---:|---:|
|
| {tbl}
|
| | **total** | | | **{tot_doc:,}** | **{tot_tok/1e6:,.1f}M** |
|
|
|
| ## Method
|
| Only **human-authored** text — no synthetic, machine-translated, or auto-transcribed
|
| data. Gates are intentionally minimal (drop short docs, non-Polish, exact duplicates,
|
| OCR garble); heavy quality filtering and mix-weighting are left to downstream training.
|
| Evaluation-set decontamination is applied/marked separately. Schema:
|
| `id, text, source, added, created, token_count, license, author`. The `license`
|
| and `author` columns are per-document metadata when upstream exposes them; older
|
| sources use the source-level license and an empty author field.
|
|
|
| ## v0.3 quality roadmap and current status
|
|
|
| The v0.2.x raw corpus is intentionally provenance-first, but its token mix is too
|
| heavy in legal/parliamentary language for natural general pretraining. The v0.3
|
| workflow therefore separates **source inclusion** from **training mix**:
|
|
|
| - cap `eurlex + parliamentary + dziennik_ustaw` to roughly **10-20%** of training
|
| tokens combined;
|
| - use source-level temperature sampling (`sqrt`, alpha `0.5`) instead of raw
|
| token-proportional sampling;
|
| - add traceably licensed contemporary/natural Polish: open web, academic prose,
|
| cultural heritage text, guides, technical documentation/blogs, Q&A, and
|
| dialogue/instruction data;
|
| - run aggressive exact, normalized, and near-duplicate removal;
|
| - reserve the final **5-15%** of training for higher-quality sources rather than
|
| the largest sources;
|
| - evaluate per-source perplexity and style contamination, not only global loss.
|
|
|
| Current v0.3 source-ingestion status:
|
|
|
| - `biblioteka_nauki`: prepared in the source registry as a direct-upstream
|
| rebuild target with per-document license and author metadata; not included in
|
| v{VERSION} parquets yet.
|
| - `europeana`: prepared in the source registry as a direct-upstream rebuild
|
| target with per-record rights statements and creator metadata; raw SpeakLeash
|
| Europeana remains excluded.
|
| - Europeana release policy: split conservatively at pre-1929 records for
|
| US-sensitive downstream reuse, and keep later/unknown records separately
|
| labeled or held until legal review.
|
| - `european_hplt_v3_pl`: WDS bins 10+9 are included in v{VERSION} after contract
|
| validation. HPLT packaging is CC0, while underlying crawled web documents can
|
| carry independent rights; downstream near-dedup and web-content review remain
|
| recommended before training.
|
|
|
| Current review artifacts:
|
|
|
| - `configs/source_candidates_v0_3.json` — candidate decisions and license policy.
|
| - `artifacts/source_license_review_v0_3.md` — source-by-source license review.
|
| - `artifacts/source_candidate_audit_v0_3.md` — generated Hugging Face metadata audit.
|
| - `artifacts/training_mix_v0_3.md` — example 1B-token training mix with legal sources capped at 15%.
|
| - `artifacts/bartek_source_ingestion_plan_2026-07-02.md` — PR contract for
|
| Biblioteka Nauki and Europeana ingestion.
|
|
|
| ## Excluded sources (transparency)
|
| Sources we reviewed and **deliberately left out** — part of the curation:
|
|
|
| | source | reason |
|
| |---|---|
|
| {excl}
|
|
|
| {DISCLAIMER}
|
| ## License & attribution
|
| Released under **CC-BY-SA-4.0** (copyleft inherited from CC-BY-SA sources such as
|
| Wikipedia/Wikisource/Wolne Lektury). Attribution due to each upstream (see datasheets)
|
| and to **SpeakLeash** as the intermediate aggregator. Retroactive-removal policy: a
|
| source that raises an objection is dropped from subsequent versions, recorded in the
|
| CHANGELOG.
|
|
|
| ## Reproduce
|
| ```bash
|
| python3 src/build_dynaword.py --all --speakleash-dir <speakleash_zst_dir> --out .
|
| python3 src/make_docs.py
|
| ```
|
| {phrase_frequency}
|
| """
|
| (ROOT / "README.md").write_text(readme)
|
|
|
|
|
|
|
| changelog_path = ROOT / "CHANGELOG.md"
|
| changelog = changelog_path.read_text() if changelog_path.exists() else "# Changelog\n"
|
| heading = f"## v{VERSION} ({RELEASE_DATE})"
|
| if heading not in changelog:
|
| entry = (
|
| f"{heading}\n\n"
|
| f"- Current release totals after parquet recount: {len(rows)} sources, "
|
| f"{tot_doc:,} docs, {tot_tok:,} tokens (tiktoken cl100k proxy).\n\n"
|
| )
|
| changelog = changelog.replace("# Changelog\n", f"# Changelog\n\n{entry}", 1)
|
| changelog_path.write_text(changelog)
|
|
|
| (ROOT / "LICENSE").write_text(
|
| "Polish DynaWord is released under Creative Commons Attribution-ShareAlike 4.0\n"
|
| "International (CC-BY-SA-4.0): https://creativecommons.org/licenses/by-sa/4.0/\n\n"
|
| "Per-source upstream licenses and attribution are documented in each\n"
|
| "data/<source>/<source>.md datasheet.\n")
|
|
|
| print(f"docs written: README + CHANGELOG + LICENSE + {len(rows)} datasheets")
|
| print(f"TOTAL {tot_doc:,} docs | {tot_tok/1e9:.2f}B tok | {tot_chr/1e9:.1f}B chars")
|
|
|
|
|
| if __name__ == "__main__":
|
| main()
|
|
|
