| | --- |
| | task_categories: |
| | - summarization |
| | language: |
| | - fr |
| | --- |
| | MLSUM-fr filtered with Aumiller et al. (1)'s library available at https://github.com/dennlinger/summaries/tree/main |
| |
|
| | 16954 samples were removed from the dataset. |
| | Breakdown by filter category: |
| | 'compression_ratio': {'train': 45, 'validation': 2, 'test': 0} samples removed across splits. |
| | 'exact_duplicate': {'train': 138, 'validation': 0, 'test': 8} samples removed across splits. |
| | 'both_duplicate': {'train': 466, 'validation': 0, 'test': 0} samples removed across splits. |
| | 'summary_duplicate': {'train': 2088, 'validation': 99, 'test': 89} samples removed across splits. |
| | 'reference_duplicate': {'train': 13843, 'validation': 119, 'test': 57} samples removed across splits. |
| | |
| | (1): Aumiller, D., Fan, J., & Gertz, M. (2023). On the State of German (Abstractive) Text Summarization. arXiv preprint arXiv:2301.07095. |