Taiwanese Hokkien to Chinese Parallel Text for Model Testing

Taiwanese Hokkien (Southern Minnan) paired with Chinese translations, compiled by Curiousfox from multiple open-source contributors. The dataset acknowledges sources including Jonshon8187/Chinese_Multi-Emotion_Dialogue_Dataset and sarahwei/Taiwanese-Minnan-Sutiau. It was last updated on March 26, 2026.

Use Cases

• Train sequence-to-sequence models to translate Taiwanese Hokkien text to Chinese text.
• Benchmark the performance of multilingual language models on Taiwanese Hokkien comprehension tasks.
• Fine-tune text encoders using the parallel sentence pairs for cross-lingual representation learning.
• Evaluate translation quality for Taiwanese Hokkien using human-verified or sourced text pairs.

Strengths

• Derived from multiple acknowledged open-source datasets, suggesting a compilation effort.
• Specifically focuses on Taiwanese Hokkien, a lower-resource Chinese dialect.

Limitations

• Unknown row count, sample size, and data volume.
• Unknown data quality metrics, such as translation accuracy or label noise.
• No column information or sample data available for inspection.

Provenance

Source

Aggregated from multiple Hugging Face datasets and andreihar.com/taibun.

Collection Method

Compilation from existing open-source text and translation datasets.

Freshness

Last updated March 26, 2026.

Geography

Likely Taiwan-centric, focusing on Taiwanese Hokkien.