Normalized Datasets for Korean LLM: 944M Documents from 42 Sources

Approximately 944.5 million normalized documents from 42 source datasets across English, Korean, Code, and Science domains, compiled by mkd-chanwoo. The dataset is formatted as JSONL and serves as a pretraining corpus for Korean language models. It was last updated on June 2, 2026.

Use Cases

• Pretraining Korean language models based on the multilingual document corpus.
• Benchmarking model performance on Korean and English text from the described domains.
• Training code generation models using the code-related documents mentioned in the description.
• Developing scientific language models based on the science domain documents.

Strengths

• Large scale with approximately 944.5 million total documents.
• Aggregates content from 42 distinct source datasets.
• Covers multiple domains including English, Korean, Code, and Science.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-06-02 01:15:05; freshness should be verified.

Provenance

Source

huggingface

Collection Method

Aggregated and normalized from 42 source datasets.

Freshness

2026-06-02 01:15:05