SlayerLab/polish-dynaword

Polish DynaWord

A continuously developed, openly-licensed, human-text Polish corpus — a Polish edition in the Dynaword family (Enevoldsen et al., arXiv:2508.02271).

v0.2.1 stable · 2,491,773 documents · 6.28B tokens (tiktoken proxy; canonical Llama-3 count at release) · 12 sources Updated: 2026-07-02

v0.3.0-preview in progress · quality/diversity remix workflow, legal-style downweighting, and filtered contemporary Polish web candidates. Biblioteka Nauki and Europeana are prepared as source-ingestion PR targets with per-document license and author metadata, but they are not part of this stable parquet release yet.

Versions

version	status	documents	tokens	notes
v0.2.1	stable release	2,491,773	6.28B	12-source stable corpus with license and author metadata columns; adds 1000_novels.
v0.2.0	previous stable	2,490,773	6.22B	Provenance-first corpus from 11 open/official sources.
v0.3.0-preview	workflow + candidate data in progress	candidate-only	TBD	Biblioteka Nauki, Europeana, and HPLT/Common Corpus style expansion pending per-source QA, dedup, and legal review.

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

• Kacper Wikieł — corpus curation, release engineering, documentation, and reproducible build workflow.
• Bart Kobyliński — source expansion work for Biblioteka Nauki and Europeana, including per-document license/author metadata collection and rebuild planning.

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
eurlex	EUR-Lex (EU legal acts, Polish)	CC-BY-4.0	243,060	2,378.1M
parliamentary	Polish Parliamentary Corpus (Sejm/Senat)	public-domain (official documents)	324,622	1,646.8M
wikisource	Polish Wikisource	CC-BY-SA-3.0	632,005	801.9M
wikipedia	Polish Wikipedia	CC-BY-SA-3.0	1,171,897	707.2M
dziennik_ustaw	Dziennik Ustaw + Monitor Polski (Polish primary legislation)	public-domain (official documents)	35,442	486.1M
wolne_lektury	Wolne Lektury (school readings)	CC-BY-SA-4.0 / Wolna Sztuka 1.3	6,141	103.0M
1000_novels	1000 Novels Corpus (CLARIN-PL)	CC-BY-4.0	1,000	60.5M
wikiquote	Polish Wikiquote (quotations)	CC-BY-SA-3.0	30,363	31.9M
eltec_pol	ELTeC-pol (European Literary Text Collection, Polish)	CC-BY-4.0	100	21.5M
wikivoyage	Polish Wikivoyage (travel guides)	CC-BY-SA-3.0	13,645	17.1M
wikibooks	Polish Wikibooks (open textbooks)	CC-BY-SA-3.0	9,112	15.6M
wikinews	Polish Wikinews	CC-BY-2.5	24,386	12.1M
total			2,491,773	6,281.9M

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 v0.2.1 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.
• ashtok897/european-hplt-v1: candidate workflow exists, but web-crawl provenance, dedup, QA, and final mix weighting are still pending before stable inclusion.

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
open_subtitles_corpus	Derivative of copyrighted film/TV dialogue; OpenSubtitles uploads largely unlicensed. Same copyright lesson as Danish Gigaword's OpenSubtitles (paper 2508.02271). Not openly licensed.
europeana_eu_pl_corpus_raw_speakleash	Aggregated items with mixed per-record rights (PD / CC-BY-NC / rights-reserved). The raw SpeakLeash redistribution is excluded; only a direct rebuild preserving per-record rights metadata may be included.
project_gutenberg_pl_corpus	Only 31 PL books (4.3MB) — PG is ~99% English; Polish PD literature already covered by wolne_lektury + wikisource (so near-redundant after dedup). Dropped to avoid the PD-in-EU per-work check (PG claims PD-in-US only) for negligible token gain.

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 k.wikiel@gmail.com — 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 k.wikiel@gmail.com; it is dropped from the next version and recorded in the CHANGELOG. Removal is retroactive going-forward (prior immutable snapshots/commits may persist).

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

python3 src/build_dynaword.py --all --speakleash-dir <speakleash_zst_dir> --out .
python3 src/make_docs.py

Results

Corpus phrase frequency (normalized by tokens)

Raw counts and token-normalized shares are regenerated from the current parquet files with src/pattern_frequency_report.py.

• Total token count (tiktoken proxy): 6,281,911,234

Pattern	Count	Share of all tokens
w roku	435,541	0.0069%
klasyfikacji	129,977	0.0021%
ustawa	587,459	0.0094%
artykuł	2,036,359	0.0324%
parlament	1,201,641	0.0191%
rozporządzenie	1,490,464	0.0237%
w pobliżu	78,823	0.0013%
mieszkańców	241,200	0.0038%
Dz.U.	939,966	0.0150%

Per-source shares

source	pattern	count	share of source tokens
1000_novels	w roku	659	0.00109%
1000_novels	klasyfikacji	14	0.00002%
1000_novels	ustawa	656	0.00108%
1000_novels	artykuł	729	0.00120%
1000_novels	parlament	240	0.00040%
1000_novels	rozporządzenie	65	0.00011%
1000_novels	w pobliżu	1,262	0.00209%
1000_novels	mieszkańców	868	0.00143%
1000_novels	Dz.U.	0	0.00000%
dziennik_ustaw	w roku	33,096	0.00681%
dziennik_ustaw	klasyfikacji	10,896	0.00224%
dziennik_ustaw	ustawa	77,376	0.01592%
dziennik_ustaw	artykuł	27,773	0.00571%
dziennik_ustaw	parlament	42,338	0.00871%
dziennik_ustaw	rozporządzenie	166,383	0.03423%
dziennik_ustaw	w pobliżu	1,087	0.00022%
dziennik_ustaw	mieszkańców	8,025	0.00165%
dziennik_ustaw	Dz.U.	158	0.00003%
eltec_pol	w roku	108	0.00050%
eltec_pol	klasyfikacji	1	0.00000%
eltec_pol	ustawa	153	0.00071%
eltec_pol	artykuł	173	0.00081%
eltec_pol	parlament	95	0.00044%
eltec_pol	rozporządzenie	35	0.00016%
eltec_pol	w pobliżu	246	0.00114%
eltec_pol	mieszkańców	214	0.00100%
eltec_pol	Dz.U.	0	0.00000%
eurlex	w roku	40,009	0.00168%
eurlex	klasyfikacji	59,428	0.00250%
eurlex	ustawa	30,368	0.00128%
eurlex	artykuł	1,774,958	0.07464%
eurlex	parlament	780,286	0.03281%
eurlex	rozporządzenie	1,202,658	0.05057%
eurlex	w pobliżu	6,088	0.00026%
eurlex	mieszkańców	9,441	0.00040%
eurlex	Dz.U.	915,707	0.03851%
parliamentary	w roku	198,192	0.01203%
parliamentary	klasyfikacji	12,637	0.00077%
parliamentary	ustawa	459,179	0.02788%
parliamentary	artykuł	182,038	0.01105%
parliamentary	parlament	309,695	0.01881%
parliamentary	rozporządzenie	113,547	0.00689%
parliamentary	w pobliżu	4,964	0.00030%
parliamentary	mieszkańców	78,048	0.00474%
parliamentary	Dz.U.	23,809	0.00145%
wikibooks	w roku	319	0.00205%
wikibooks	klasyfikacji	37	0.00024%
wikibooks	ustawa	165	0.00106%
wikibooks	artykuł	732	0.00470%
wikibooks	parlament	283	0.00182%
wikibooks	rozporządzenie	131	0.00084%
wikibooks	w pobliżu	125	0.00080%
wikibooks	mieszkańców	204	0.00131%
wikibooks	Dz.U.	16	0.00010%
wikinews	w roku	449	0.00370%
wikinews	klasyfikacji	639	0.00526%
wikinews	ustawa	407	0.00335%
wikinews	artykuł	2,474	0.02038%
wikinews	parlament	2,530	0.02084%
wikinews	rozporządzenie	168	0.00138%
wikinews	w pobliżu	455	0.00375%
wikinews	mieszkańców	1,014	0.00835%
wikinews	Dz.U.	27	0.00022%
wikipedia	w roku	143,023	0.02022%
wikipedia	klasyfikacji	46,043	0.00651%
wikipedia	ustawa	9,536	0.00135%
wikipedia	artykuł	28,165	0.00398%
wikipedia	parlament	57,863	0.00818%
wikipedia	rozporządzenie	5,637	0.00080%
wikipedia	w pobliżu	41,915	0.00593%
wikipedia	mieszkańców	122,766	0.01736%
wikipedia	Dz.U.	232	0.00003%
wikiquote	w roku	608	0.00191%
wikiquote	klasyfikacji	15	0.00005%
wikiquote	ustawa	357	0.00112%
wikiquote	artykuł	606	0.00190%
wikiquote	parlament	1,271	0.00398%
wikiquote	rozporządzenie	27	0.00008%
wikiquote	w pobliżu	207	0.00065%
wikiquote	mieszkańców	527	0.00165%
wikiquote	Dz.U.	6	0.00002%
wikisource	w roku	16,571	0.00207%
wikisource	klasyfikacji	166	0.00002%
wikisource	ustawa	8,156	0.00102%
wikisource	artykuł	16,230	0.00202%
wikisource	parlament	6,119	0.00076%
wikisource	rozporządzenie	1,651	0.00021%
wikisource	w pobliżu	13,921	0.00174%
wikisource	mieszkańców	14,335	0.00179%
wikisource	Dz.U.	5	0.00000%
wikivoyage	w roku	609	0.00356%
wikivoyage	klasyfikacji	21	0.00012%
wikivoyage	ustawa	46	0.00027%
wikivoyage	artykuł	657	0.00384%
wikivoyage	parlament	249	0.00145%
wikivoyage	rozporządzenie	34	0.00020%
wikivoyage	w pobliżu	6,480	0.03783%
wikivoyage	mieszkańców	4,100	0.02394%
wikivoyage	Dz.U.	1	0.00001%
wolne_lektury	w roku	1,898	0.00184%
wolne_lektury	klasyfikacji	80	0.00008%
wolne_lektury	ustawa	1,060	0.00103%
wolne_lektury	artykuł	1,824	0.00177%
wolne_lektury	parlament	672	0.00065%
wolne_lektury	rozporządzenie	128	0.00012%
wolne_lektury	w pobliżu	2,073	0.00201%
wolne_lektury	mieszkańców	1,658	0.00161%
wolne_lektury	Dz.U.	5	0.00000%