Simplified Chinese Wikipedia Articles for Language Model Pretraining

yuhuanstudio's Wikipedia Pretrain Zh dataset provides text from the Chinese Wikipedia, converted to Simplified Chinese. The data appears to be structured as JSON objects with 'title' and 'text' fields, containing article excerpts. The dataset was last updated on 2026-06-02.

Use Cases

• Pretraining large language models on Chinese text based on the provided Wikipedia article excerpts.
• Fine-tuning models for Chinese question-answering based on factual text content.
• Training text generation models on structured, encyclopedic Chinese prose.
• Analyzing linguistic patterns in formal, Simplified Chinese writing.

Strengths

• Text is processed to use Simplified Chinese exclusively, avoiding mixed Simplified/Traditional characters.
• Data is structured with clear 'title' and 'text' fields for each entry, as shown in the example.

Limitations

• Description metadata is limited; actual data quality, size, and completeness require manual inspection after download.
• Row count, total size, and license information are unknown, which may limit suitability assessment.
• Column-level documentation beyond the two provided fields is absent; field semantics must be inferred after download.

Provenance

Source

Chinese Wikipedia (zh.wikipedia.org)

Collection Method

Monthly dumps from Wikipedia, converted to Simplified Chinese using OpenCC.

Freshness

Last updated 2026-06-02 13:25:45; freshness should be verified.