| dataset_info: | |
| features: | |
| - name: transcript | |
| dtype: string | |
| splits: | |
| - name: train | |
| num_bytes: 287264818 | |
| num_examples: 261469 | |
| download_size: 176236194 | |
| dataset_size: 287264818 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-* | |
| This dataset consists of a mixture of [TED talk transcripts](https://www.kaggle.com/datasets/miguelcorraljr/ted-ultimate-dataset), the high-quality [Aeon essays](https://www.kaggle.com/datasets/mannacharya/aeon-essays-dataset), as well as a filtered selection of [NPR Interview Transcripts](https://www.kaggle.com/datasets/shuyangli94/interview-npr-media-dialog-transcripts). | |
| All NPR transcripts were filtered to contain only the turns with > 100 words. |