CNN DailyMail Summarization Benchmark for Model Evaluation

Autoevaluate hosts a dataset for evaluating summarization models, derived from the CNN/DailyMail news article dataset. The project was created by the 'autoevaluate' organization and last updated on July 27, 2022. It is intended for staging and benchmarking automated evaluation metrics.

Use Cases

• Benchmarking abstractive summarization models using the original article text and reference summaries.
• Training automated evaluation metrics by comparing model-generated summaries to human-written highlights.
• Analyzing model performance on specific data splits or subsets identified by the project ID '5C70C9C5 11935598'.

Strengths

• Based on the established CNN/DailyMail dataset, a standard benchmark for summarization.
• Provides a structured evaluation project with a specific identifier for reproducible testing.

Limitations

• Specific row count, size, and modifications from the original dataset are unknown.
• The dataset's recency is limited, with the last update in mid-2022.

Provenance

Source

Derived from the CNN/DailyMail dataset.

Collection Method

Curated for automated evaluation staging by autoevaluate.

Freshness

Last updated 2022-07-27.