Indonesian Wikipedia Summarization Dataset

Dataset contains 10.000 Indonesian Wikipedia article summarization pairs. Each entry has an instruction (input) and a summary (output) in Indonesian. Designed for training/evaluating Indonesian LLMs.

Use Cases

• Train sequence-to-sequence models
• evaluate summarization accuracy
• Fine-tune pre-trained Indonesian models

Strengths

• 10.000 pairs provide substantial training data
• Clean instruction-output format suitable for instruction tuning
• Focus on Indonesian fills a language gap

Limitations

• Summaries may vary in quality as they are auto-generated
• No column-level documentation available
• May require preprocessing for model ingestion

Provenance

Source

Wikipedia

Collection Method

Auto-generated from articles

Time Range

unknown

Freshness

Last updated 2026-05-09 16:29 :36

Geography

Indonesia