French Wikipedia Preprocessed for Language Model Training

French Wikipedia articles preprocessed for machine learning. The dataset contains 11.0 GB of cleaned text from the French Wikipedia dump of May 2026. It was processed by raj2708 and is licensed under CC-BY-SA 4.0.

Use Cases

• Pretrain large language models based on the cleaned French text corpus.
• Fine-tune models for French language understanding tasks based on the preprocessed articles.
• Analyze linguistic patterns and knowledge representation in French Wikipedia.
• Benchmark text generation models on French language tasks.

Strengths

• 11.0 GB of cleaned text provides a substantial corpus for training.
• Explicit preprocessing steps include parsing wikitext to plain text and filtering redirects and short articles.
• Clear licensing (CC-BY-SA 4.0) and source attribution (fr.wikipedia.org).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified as the dump date is listed as May 2026.

Provenance

Source

fr.wikipedia.org

Collection Method

Downloaded full French Wikipedia XML BZ2 dump, parsed with mwparserfromhell, filtered and cleaned.

Time Range

Snapshot from May 2026.

Freshness

Dump date is May 2026; last metadata update was 2026-06-01.