Modern Chinese to Lu Xun Style Dataset with 7,000 Text Pairs

7,000 Chinese text pairs for modern Chinese to Lu Xun style rewriting. The modern Chinese source side was generated by DeepSeek V4 Flash through an API-based modernization pipeline, while the target side contains Lu Xun style Chinese text. The dataset was created by liuyanliang and last updated on Hugging Face in June 2026.

Use Cases

• Fine-tuning text generation models for Chinese style transfer based on the described text-to-text task.
• Training models to rewrite modern Chinese into classical literary styles based on the Lu Xun style target text.
• Benchmarking the performance of language models on Chinese stylistic rewriting tasks.
• Studying the linguistic features of Lu Xun's writing style through machine learning.

Strengths

• Contains 7,000 text pairs, providing a substantial corpus for training.
• Source text was generated using a specific, named model (DeepSeek V4 Flash) via an API pipeline.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-06-14 07:56:39; freshness should be verified.

Provenance

Source

Hugging Face dataset by liuyanliang.

Collection Method

Modern Chinese source generated by DeepSeek V4 Flash via an API pipeline; target side is Lu Xun style text.

Freshness

Last updated 2026-06-14 07:56:39.