News Summarization VI: 75,361 Vietnamese Article-Summary Pairs for NLP

75,361 Vietnamese news articles with corresponding summaries, split into 65,361 training and 10,000 test examples. The dataset was created by vietgpt and last updated on the Hugging Face platform in July 2023. Summaries were generated using the text-davinci-003 model.

Use Cases

• Fine-tuning abstractive summarization models based on the article-summary pairs.
• Benchmarking the performance of large language models on Vietnamese text generation tasks.
• Training sequence-to-sequence models for content condensation in news media.
• Studying the characteristics of machine-generated summaries for Vietnamese text.

Strengths

• Contains 75,361 total examples, providing substantial data for model training.
• Includes a predefined 10,000-example test set for standardized evaluation.
• Summaries were generated by a high-capacity model (text-davinci-003), suggesting consistent formatting.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the full dataset is confirmed, but the original source and collection methodology details are limited.
• Data may reflect temporal or source bias inherent to the original news corpus used for generation.

Provenance

Source

https://github.com/binhvq/news-corpus

Collection Method

Summaries were generated by the text-davinci-003 model based on articles from a news corpus.

Time Range

null

Freshness

Last updated 2023-07-04 05:30:39; freshness should be verified.

Geography

null