Model Predictions for Summarization on CNN DailyMail Test Set

Comprising model predictions for a summarization task on the CNN/DailyMail dataset's test split. The predictions were generated by the pszemraj/long-t5-tglobal-base-16384-book-summary model using the AutoTrain evaluator in July 2022.

Use Cases

• Compare predicted summaries from the long-t5-tglobal-base-16384-book-summary model against reference summaries for the CNN/DailyMail test split.
• Analyze the performance of the AutoTrain evaluator on the summarization task for the specified model and dataset.
• Use the predictions as a baseline for evaluating new summarization models on the CNN/DailyMail 3.0.0 configuration.

Strengths

• Predictions are tied to a specific, documented model architecture (long-t5-tglobal-base-16384-book-summary).
• Evaluated on a standard benchmark dataset (CNN/DailyMail) using its test split and version 3.0.0 configuration.
• Generated by an automated evaluation platform (AutoTrain/Hugging Face).

Limitations

• The dataset's size, row count, column structure, and sample data are unknown.
• The data is from mid-2022 and may not reflect the current performance of the model or evaluator.
• Without column details, the specific format of predictions (e.g., generated text, scores) is unclear.

Provenance

Source

Hugging Face AutoTrain Evaluator.

Collection Method

Automated model evaluation generating predictions for a summarization task.

Time Range

Evaluation timestamped 2022-07-16.

Freshness

Last updated 2022-07-16.