cc-re-2020: pre-ChatGPT human-written English web text

5,000,000 documents of clean English, overwhelmingly human-written web text, extracted from the 9 CommonCrawl crawls of 2020, globally deduplicated within the year (and sharing no identical document with the sister year), annotated with WebOrganizer topic × format labels, and selected so the per-format and per-topic row totals are each as uniform as the eligible pool allows (water-filled marginal targets + exact LP; scarce labels contribute their entire pool — see reports/selection_report.json achieved_marginals for the realized totals).

This corpus is the "human" class for training and evaluating LLM-text detectors. Two properties make it suitable:

1. Temporal purity. Every page was crawled in 2020 — before the release of ChatGPT (Nov 2022) and before large-scale LLM text existed on the public web — which rules out chatbot-era AI contamination by crawl date. Pre-LLM machine text (synonym-spinner spam, raw machine translation, template/doorway pages) does exist on the 2020 web; after remediation a manual audit measures its residual at ~5% of rows (95% CI 0.7–9.3; see Known limitations).
2. No LLM in the loop. No LLM ever generated, rewrote, paraphrased, or summarized any text in this dataset. Models are used read-only: language ID, fastText quality scoring, and topic/format classification. The text bytes are what trafilatura extracted from the original 2020 HTML, post-processed only by (a) FineWeb's standard heuristic filters (the C4 rules also drop boilerplate lines — cookie/policy notices, javascript mentions, sub-3-word lines — and strip bracketed citation markers, exactly as in FineWeb), and (b) a deterministic, removal-only line scrubber for page furniture (cookie banners, share widgets, bylines/postmeta, breadcrumbs, pagination, footer notices — 18 documented rules, no model involved; boiler_lines_removed records the per-row count, ~5.6% of rows touched, ~0.06% of characters). No masking or any other rewriting is applied. Some navigation/UI residue still survives in a minority of rows — deterministic scrubbing is precision-first, not exhaustive.

Sister dataset: the corresponding corpus for the other pilot year at G-reen/cc-re-2020 / G-reen/cc-re-2021.

Pipeline

The pipeline composes the published recipes of FineWeb, GneissWeb, Ultra-FineWeb, and WebOrganizer (extraction → filtering → dedup → annotation → balanced selection). Full source code: the ccre pipeline repo (see reports/pipeline_config.yaml in this repo for the exact configuration used).

1. Manifest — enumerate the year's CommonCrawl dumps; sample WARC files uniformly across crawls and segments (seeded, reproducible).
2. Extraction + heuristic filtering (exact FineWeb stage-A recipe, via datatrove): WARC records (not WET) → UT1 adult/URL blocklist → trafilatura (favour_precision=True, no comments) → fastText lid.176 English ≥ 0.65 → Gopher repetition filter → Gopher quality filter → C4 filters (without the terminal-punctuation line rule) → FineWeb quality filters.
3. Cheap junk fences (from GneissWeb, applied as lenient floors): extreme-tokenized-document filter 0.1 < tokens/char < 0.5 (catches mixed-script leakage, encoded junk, and permutation spam that language ID misses); McAlpine-EFLAW readability is stored as metadata only — never used as a filter, to preserve register diversity.
4. Quality floors, not ceilings (Ultra-FineWeb + DCLM): every document is scored by the Ultra-FineWeb-en fastText classifier (with its exact DeepSeek-V2-token preprocessing) and the DCLM fastText. Keep rule: dclm_score ≥ 0.002 (GneissWeb's permissive floor); the UFW score is recorded as metadata only (no UFW floor is applied — a UFW gate would skew the corpus toward educational text). These thresholds are deliberately permissive floors: the goal is diverse, representative human writing, not FineWeb-edu-style "educational" ceilings that would collapse the register distribution a detector must learn.
5. Global per-year MinHash deduplication (datatrove 4-stage MinHash: signatures → buckets → clusters → filter), sha1 64-bit, 14 buckets × 8 hashes over word 5-grams (≈ 0.72 Jaccard threshold), across all of the year's crawls (unlike FineWeb's per-dump dedup — uniqueness matters more than benchmark-optimal mixing for detector data). Duplicate-cluster sizes are retained as metadata.
6. Topic × format annotation (WebOrganizer): the URL-aware TopicClassifier and FormatClassifier (24 labels each, gte-base-en-v1.5 backbone, max_length 512) label every document; softmax confidences are stored.
7. Marginal-uniform selection — the per-FORMAT and per-TOPIC row totals are equalized as far as pool supply allows: water-filled marginal targets per axis (labels whose entire eligible pool is below the common level contribute everything they have; all other labels get an equal share), reconciled at the topic × format cell level by an exact LP (HiGHS) minimizing total L1 deviation from both axes' targets subject to per-cell pool capacities, down to 5,000,000 rows; within-cell choice is random (seeded). Achieved marginals ship in reports/selection_report.json (achieved_marginals). Note the marginals are as-uniform-as-feasible, not exactly uniform: scarce labels are supply-limited, and residual joint infeasibility (a label's supply concentrated in saturated labels of the other axis) is split across axes by the LP. Before allocation, eligibility filters exclude non-prose and corrupted rows: dropped topics ['Adult'], dropped formats ['Truncated', 'Content Listing', 'Structured Data'], language_score >= 0.925, and tokens_per_char in [0.13, 0.3]. Formats in {'Spam / Ads': 0.01} are deterministically downsampled to the given keep-fraction (the Spam / Ads label was measured ~96% machine-generated junk in a manual audit; a 1% remnant is retained deliberately). Cross-year hygiene: rows whose exact text appears in the other published cc-re year are excluded per cross_year_exclude ({2021: 2020}), so the two years share no identical documents and can serve as leakage-free train/val splits. Seeded and reproducible; see reports/selection_report.json for the full funnel, filter exclusion counts, and per-cell counts.

Schema

column	type	description
boiler_lines_removed	int32	Lines removed by the deterministic page-furniture scrubber (0 = untouched)
crawl	string	CommonCrawl dump id, e.g. CC-MAIN-2020-24
date	string	Page fetch time (WARC-Date, ISO 8601)
dclm_score	float32	DCLM fastText (fasttext-oh-eli5) __label__hq probability
eflaw_readability	—
format	string	WebOrganizer format label (24-way)
format_prob	float32	Format classifier softmax confidence
id	string	Stable unique id: <crawl>/<warc-file#record>
language_score	float32	fastText lid.176 P(en); kept only if ≥ 0.65
n_emails	int32	Count of email addresses in text (FineWeb PII regex); text is NOT masked
n_ips	int32	Count of public IPv4 addresses in text (FineWeb PII regex); text is NOT masked
text	string	Extracted clean text (trafilatura; never LLM-generated or rewritten)
tokens_per_char	float32	GneissWeb extreme-token stat: tokens / chars (GPT-2)
topic	string	WebOrganizer topic label (24-way)
topic_prob	float32	Topic classifier softmax confidence
ufw_score	float32	Ultra-FineWeb-en fastText quality score (P(pos))
url	string	Source page URL (WARC-Target-URI)
word_count	int32	Whitespace word count

Statistics

Overview

Statistic	Value
Rows	5,000,000
Parquet shards	10
Total words	3,714,567,900
Total text characters	21,743,337,692
Crawls	9
Fetch-date range	2020-01-17T12:34:15Z → 2020-12-06T03:19:19Z

Distributions

Metric	Mean	Median	p5	p95	Min	Max
Words per document	742.9	465	102	2,122	50	98,409
Characters per document	4,349	2,720	590	12,417	211	673,592
Ultra-FineWeb quality score	0.1474	0.02878	0.0003769	0.8041	0	1

Rows per crawl

Crawl	Rows	Share
CC-MAIN-2020-05	709,760	14.20%
CC-MAIN-2020-10	530,604	10.61%
CC-MAIN-2020-16	646,850	12.94%
CC-MAIN-2020-24	537,027	10.74%
CC-MAIN-2020-29	660,864	13.22%
CC-MAIN-2020-34	500,004	10.00%
CC-MAIN-2020-40	533,521	10.67%
CC-MAIN-2020-45	447,044	8.94%
CC-MAIN-2020-50	434,326	8.69%

Topics

Topic	Rows	Share
Health	303,967	6.08%
Finance & Business	303,617	6.07%
Religion	245,933	4.92%
Crime & Law	245,749	4.91%
Literature	245,735	4.91%
Science & Tech.	245,699	4.91%
Education & Jobs	245,681	4.91%
Food & Dining	245,679	4.91%
Politics	245,656	4.91%
Home & Hobbies	245,622	4.91%
Entertainment	245,590	4.91%
Sports & Fitness	245,520	4.91%
Transportation	237,060	4.74%
Art & Design	219,611	4.39%
Games	207,822	4.16%
Social Life	195,494	3.91%
Travel	187,698	3.75%
Software	184,511	3.69%
History	165,207	3.30%
Fashion & Beauty	164,410	3.29%
Industrial	139,137	2.78%
Hardware	132,302	2.65%
Software Dev.	102,300	2.05%

Formats

Format	Rows	Share
Personal Blog	539,660	10.79%
Product Page	481,593	9.63%
Knowledge Article	481,543	9.63%
Nonfiction Writing	481,471	9.63%
News Article	481,355	9.63%
Tutorial	466,245	9.32%
Listicle	322,188	6.44%
Comment Section	288,972	5.78%
News (Org.)	252,187	5.04%
About (Org.)	188,425	3.77%
User Review	179,403	3.59%
Audio Transcript	171,848	3.44%
Creative Writing	135,074	2.70%
Academic Writing	125,711	2.51%
About (Pers.)	121,295	2.43%
Customer Support	93,879	1.88%
Q&A Forum	82,655	1.65%
FAQ	57,929	1.16%
Legal Notices	36,046	0.72%
Documentation	11,117	0.22%
Spam / Ads	1,404	0.03%

Topic × format cells

Non-empty cells: 462 (of 24 × 24 = 576 possible). Largest cell: Health × Knowledge Article (94,735 rows); smallest non-empty cell: Social Life × Documentation (2 rows).

Top 30 cells:

Topic	Format	Rows
Health	Knowledge Article	94,735
Crime & Law	News Article	93,456
Sports & Fitness	News Article	81,265
Home & Hobbies	Tutorial	76,753
Food & Dining	Personal Blog	76,747
Religion	Personal Blog	72,031
Literature	Creative Writing	71,517
Finance & Business	Tutorial	71,238
Politics	Nonfiction Writing	67,308
Social Life	Personal Blog	61,049
Art & Design	Personal Blog	58,249
Science & Tech.	News Article	57,356
Politics	News Article	55,797
Fashion & Beauty	Product Page	54,334
History	Knowledge Article	50,671
Transportation	News Article	49,793
Health	Tutorial	48,223
Travel	Product Page	47,993
Food & Dining	Tutorial	45,304
Finance & Business	Knowledge Article	45,272
Art & Design	Product Page	44,555
Education & Jobs	Tutorial	44,310
Health	Listicle	43,876
Entertainment	Nonfiction Writing	43,311
Literature	Product Page	42,519
Transportation	Product Page	42,008
Entertainment	User Review	41,637
Finance & Business	News (Org.)	41,477
Religion	Nonfiction Writing	41,298
Industrial	Product Page	39,996

Known limitations (measured, from manual audits)

Rates from hand-read random samples (n=100–120) and full-corpus scans; details in reports/audit_post_remediation.md:

Limitation	Measured rate	Note
Pre-LLM machine text (spinners, raw MT, doorway/template pages)	~5% of rows (CI 0.7–9.3)	not concentrated in any filterable label
Content-farm / SEO filler (human-typed, formulaic)	~11%	possible home of undetectable pre-2022 neural text
Page-furniture residue (bylines, widgets, notices)	~1 in 4 rows has some	scrubber is precision-first, not exhaustive
Extraction damage (pipe residue, dropped lists, merged words)	2.8% / ~9% / ~10%	inherent to HTML→text; not remediable without re-extraction
Scraped/republished third-party content	~12%	url/date are not reliable provenance for these
Scrubber over-removal (real lines lost)	~0.1% of rows	rows with boiler_lines_removed >= 2 are the risk set
Low-confidence format labels	format_prob < 0.5 on ~10%	filter on format_prob before conditioning on format

Also note: word_count and all scores (language_score, ufw_score, dclm_score, tokens_per_char, eflaw_readability) were computed on the pre-scrub text; rows with boiler_lines_removed > 0 (~5.5%) are slightly shorter than word_count claims.

Provenance, limitations, and bias

• Source: CommonCrawl 2020 crawls, fetched from https://data.commoncrawl.org. Use of the underlying data is subject to the CommonCrawl Terms of Use.
• Extraction bias: text is trafilatura-extracted with favour_precision=True; most boilerplate, navigation, and comments are removed (extraction + C4 line rules + the deterministic scrubber), but a minority of rows retain some page-furniture residue, and extraction can damage formatting (lost lists, table→pipe residue). Detectors trained on this corpus should compare against AI text formatted the same way (plain extracted text, no HTML).
• Selection bias: English only (lid.176 ≥ 0.65), quality floors as above, and topic × format rebalancing — the corpus is not a natural sample of the 2020 web. Cells that are near-empty on the real web (e.g. Adult × Academic Writing) remain small; see the cell table above and reports/selection_report.json.
• Residual contamination: pre-ChatGPT does not mean zero machine text — 2020 web pages can contain template-generated, machine-translated, or spun/SEO text. The quality floors and junk fences remove most of it, but no guarantee is absolute.
• Content warning: despite URL blocklists and quality filtering, web text may contain offensive, inaccurate, or otherwise objectionable content, and may mention personal data present on public 2020 web pages.
• PII is deliberately NOT masked. This corpus exists to train detectors that distinguish human from AI text; masking emails/IPs with stock replacement strings (as FineWeb does) would inject an "anonymization artifact" that appears only in the human class — a spurious, learnable shortcut — and would make the text something no human actually wrote. The text is redistributed exactly as it appeared on the public 2020 web (as in CommonCrawl itself). Per-row n_emails and n_ips columns count FineWeb-definition PII hits (email regex; public IPv4) so downstream users can filter or apply their own masking if their use case requires it.
• Takedown: to report content (including personal data) for removal, open a discussion on this repo.

License

The curation code and this compilation are released under ODC-By 1.0 (like FineWeb); use of the underlying content remains subject to the CommonCrawl Terms of Use.

Citations

This dataset composes methods from the following works — please cite them when using it:

@inproceedings{penedo2024fineweb,
  title  = {The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale},
  author = {Penedo, Guilherme and Kydl{\'i}{\v{c}}ek, Hynek and Ben allal, Loubna and Lozhkov, Anton and Mitchell, Margaret and Raffel, Colin and Von Werra, Leandro and Wolf, Thomas},
  booktitle = {NeurIPS Datasets and Benchmarks},
  year   = {2024},
  url    = {https://arxiv.org/abs/2406.17557}
}

@article{wang2025ultrafineweb,
  title  = {Ultra-FineWeb: Efficient Data Filtering and Verification for High-Quality LLM Training Data},
  author = {Wang, Yudong and Sun, Zixuan and others},
  journal = {arXiv preprint arXiv:2505.05427},
  year   = {2025},
  url    = {https://arxiv.org/abs/2505.05427}
}

@article{gneissweb2025,
  title  = {GneissWeb: Preparing High Quality Data for LLMs at Scale},
  author = {Gohari, Hajar Emami and Kadhe, Swanand Ravindra and others},
  journal = {arXiv preprint arXiv:2502.14907},
  year   = {2025},
  url    = {https://arxiv.org/abs/2502.14907}
}

@inproceedings{wettig2025weborganizer,
  title  = {Organize the Web: Constructing Domains Enhances Pre-Training Data Curation},
  author = {Wettig, Alexander and Lo, Kyle and Min, Sewon and Hajishirzi, Hannaneh and Chen, Danqi and Soldaini, Luca},
  booktitle = {ICML},
  year   = {2025},
  url    = {https://arxiv.org/abs/2502.10341}
}

Built with datatrove. Data from CommonCrawl. Published 2026-08-31.