open-index/open-markdown

Open Markdown

Clean markdown from the web, ready for training and retrieval

What is it?

Open Markdown is a large-scale web text dataset built from Common Crawl. Common Crawl is a non-profit that crawls the web and freely provides its archives and datasets to the public — see their latest crawl announcement for details on the source data. Every page goes through a pipeline that extracts the main content from raw HTML, converts it to clean Markdown, and packages the result into Parquet files with useful WARC metadata for traceability.

The dataset currently includes crawl CC-MAIN-2026-12 with 348,605,195 documents across 20252 shards. Processed 45.2 TB of raw HTML into 2.9 TB of clean Markdown — a 96.9% reduction. We plan to add more snapshots over time.

Live Progress

Processing at 116.0 shards/hour — 20,252 of 100,000 done (20.25%)

Estimated completion: April 25, 2026 (29 days)

Current server: 6 CPU cores, 12 GB RAM (6.4 GB available), 103 GB disk free

Memory per session: avg 580 MB, peak 799 MB (measured via VmRSS)

With 10 identical servers: 1160 shards/hour → March 31, 2026 (3 days)

Open Markdown is released under the Open Data Commons Attribution License (ODC-By) v1.0, the same license used by Common Crawl.

What is being released?

Each Common Crawl WARC file (~1 GB of compressed HTML) becomes one Parquet shard. The shards live under a crawl-specific directory so multiple snapshots can coexist:

data/
  CC-MAIN-2026-12/
    00/
      00/
        000000.parquet
        000001.parquet
        ...
      01/
        000100.parquet
        ...
    01/
      ...

Every row in a Parquet file is one web page. Each row includes the warc_record_id and warc_refers_to fields parsed from the original WARC headers, so you can trace any document back to its source record. We also store html_length and markdown_length to measure the compression from raw HTML to clean markdown.

How to download and use Open Markdown

Using datasets

from datasets import load_dataset

# stream the entire dataset
ds = load_dataset("open-index/open-markdown", name="CC-MAIN-2026-12", split="train", streaming=True)
for doc in ds:
    print(doc["url"], len(doc["markdown"]))

# load a single shard into memory
ds = load_dataset(
    "open-index/open-markdown",
    data_files="data/CC-MAIN-2026-12/00/00/000000.parquet",
    split="train",
)

Using huggingface_hub

from huggingface_hub import snapshot_download

folder = snapshot_download(
    "open-index/open-markdown",
    repo_type="dataset",
    local_dir="./open-index/",
    allow_patterns="data/CC-MAIN-2026-12/**/*.parquet",
)

For faster downloads, install pip install huggingface_hub[hf_transfer] and set HF_HUB_ENABLE_HF_TRANSFER=1.

Using DuckDB

SELECT url, host, markdown_length
FROM read_parquet('hf://datasets/open-index/open-markdown/data/CC-MAIN-2026-12/**/*.parquet')
WHERE host = 'en.wikipedia.org'
LIMIT 10;

Dataset card for Open Markdown

Dataset Structure

Data Instance

The following is an example row from the dataset:

{
  "doc_id": "6aaa5be7-a917-5105-aa60-e39ea1d087fc",
  "url": "https://example.com/article/interesting-topic",
  "host": "example.com",
  "crawl_date": "2026-02-06T18:14:58Z",
  "warc_record_id": "<urn:uuid:a1b2c3d4-e5f6-7890-abcd-ef1234567890>",
  "warc_refers_to": "<urn:uuid:f9e8d7c6-b5a4-3210-fedc-ba0987654321>",
  "html_length": 48210,
  "markdown_length": 3847,
  "markdown": "# Interesting Topic\n\nThis is the main content of the page..."
}

Data Fields

Column	Type	Description
doc_id	string	Deterministic UUID v5 derived from the canonical URL: doc_id = UUID5(NamespaceURL, url) — identical URLs always produce the same doc_id across crawls
url	string	Original URL of the crawled page
host	string	Lowercase hostname extracted from the URL
crawl_date	string	RFC 3339 timestamp from the WARC record
warc_record_id	string	Full WARC-Record-ID of this conversion record (<urn:uuid:...>)
warc_refers_to	string	WARC-Record-ID of the original HTTP response this was converted from
html_length	int64	Byte length of the original HTML body before conversion
markdown_length	int64	Byte length of the converted markdown body
markdown	string	Clean markdown content extracted from the page

Data Splits

The default subset includes all available data across all crawl snapshots. You can also load a specific crawl by using its ID as the config name (e.g. CC-MAIN-2026-12).

Dataset Creation

Curation Rationale

Most open web datasets either release raw text without structure or keep the HTML and leave parsing to the user. Open Markdown sits in between: it converts every page to Markdown so the content is immediately usable for training, while preserving key WARC identifiers (warc_record_id, warc_refers_to) so you can always trace back to the source record.

Source Data

The source data consists of web pages crawled by the Common Crawl foundation. Common Crawl archives billions of pages across the public web and makes the raw WARC files freely available on Amazon S3.

Data Processing Steps

The processing pipeline runs as a single-pass direct conversion:

1. Download raw .warc.gz files from Common Crawl S3 (each file is roughly 1 GB compressed)
2. Filter to keep only HTTP 200 responses with a text/html content type, discarding images, scripts, redirects, and error pages
3. Convert HTML to clean Markdown using a lightweight tokenizer-based extractor that strips tags, scripts, styles, navigation, and boilerplate — keeping only the main content
4. Export directly to Apache Parquet with Zstd compression, 100,000 rows per row group

No intermediate files are created — the pipeline streams from compressed WARC through conversion directly into Parquet. Pages that produce empty conversions are dropped.

Compression Ratios

Numbers below are actual measurements summed across all 20252 files of CC-MAIN-2026-12 (348,605,195 pages total), projected to the full crawl of 100,000 WARC files.

Stage	20252 files (measured)	100,000 files (projected)	Reduction
Raw WARC (.warc.gz, downloaded)	~16.0 TB	~79.2 TB	—
HTML extracted (uncompressed)	45.2 TB	~223.1 TB	—
Markdown (clean text)	2.9 TB	~6.8 TB	-96.9% vs HTML
Final Parquet (Zstd)	934.5 GB	~4.5 TB	-68.9% vs markdown

The big win is HTML → Markdown conversion: the tokenizer strips all tags, scripts, styles, navigation, and ads, keeping only the main content. This cuts 45.2 TB of uncompressed HTML down to 2.9 TB of markdown — a 96.9% reduction. Parquet with Zstd then compresses the markdown a further 68.9%.

End to end: ~16.0 TB of raw gzipped WARCs becomes 934.5 GB of Parquet — a 94.3% total reduction — containing 348,605,195 clean markdown documents.

Processing Times

Pipeline timings across 20252 shards of CC-MAIN-2026-12:

Download (raw WARC)                   ██████████░░░░░░░░░░░░░░  217h 51m 32s
Convert  (HTML → Markdown → Parquet)  ████████████████████████  506h 3m 23s
Publish  (HuggingFace)                ███░░░░░░░░░░░░░░░░░░░░░  74h 40m 45s

Dataset Charts

Personal and Sensitive Information

No additional PII filtering is applied beyond what Common Crawl provides. As the dataset is sourced from the public web, it is likely that some personally identifiable information is present. If you find your own PII in the dataset and would like it removed, please open an issue on the repository.

Considerations for Using the Data

Social Impact

By releasing both the dataset and the full processing pipeline, we aim to lower the barrier to training and evaluating language models on high quality web data. Researchers and practitioners who cannot afford to run their own Common Crawl processing pipelines can use Open Markdown directly.

Discussion of Biases

Open Markdown inherits the biases present in Common Crawl and the public web at large. The trafilatura extraction step favors article-like pages and may underrepresent content from forums, social media, and non-standard page layouts. We have not applied any machine-learning-based quality or toxicity filters, as such filters have been shown to disproportionately remove content from certain dialects and communities.

Known Limitations

Code-heavy pages may not convert well to Markdown. If you are training a model that needs strong code performance, consider supplementing Open Markdown with a dedicated code dataset such as The Stack v2. Similarly, highly structured pages like Wikipedia may have better formatting in dedicated Wikipedia dumps than in their Common Crawl versions.

Additional Information

Licensing

The dataset is released under the Open Data Commons Attribution License (ODC-By) v1.0. The use of this dataset is also subject to Common Crawl's Terms of Use. The original content remains subject to the rights and terms of its respective publishers.

Contact

Please open a discussion on the Community tab for questions, feedback, or issues.