AG News Dataset For Model Evaluation

An evaluation dataset from the autoevaluate platform, likely derived from the AG News corpus for testing text classification models. The project was created by 'autoevaluate' and last updated in July 2022. Specific details on row count, columns, and exact content are unavailable from the provided metadata.

Use Cases

• Benchmarking model accuracy on the AG News topic classification task using the standard label categories.
• Evaluating model robustness by testing on a specific split or subset of the AG News data referenced by the project ID.
• Comparing text classification algorithms using the predefined news article features and target labels.

Strengths

• Based on the established AG News benchmark corpus for text classification.
• Project identifier and timestamp provide traceability to a specific evaluation run.

Limitations

• No metadata on dataset size, columns, or specific modifications is provided, limiting reproducibility.
• The dataset appears to be a static snapshot from mid-2022, which may not reflect current model performance on newer data.

Provenance

Source

autoevaluate platform, likely derived from the AG News dataset.

Collection Method

Presumably a processed or sampled version created for model evaluation staging.

Time Range

null

Freshness

Last updated in July 2022; no update frequency indicated.

Geography

null