Datasets:
Improve dataset card: Add paper/code links, task categories, and sample usage
#1
by
nielsr
HF Staff
- opened
This PR improves the dataset card for the repro-rl-data by:
- Adding a link to the associated paper: RePro: Training Language Models to Faithfully Recycle the Web for Pretraining
- Adding a link to the GitHub repository: https://github.com/cxcscmu/RePro
- Adding
task_categories: ['text-generation', 'reinforcement-learning']to the metadata for improved discoverability, reflecting the dataset's use in training an LM with RL for text rephrasing. - Including a "Sample Usage" section with a code snippet from the GitHub README, demonstrating how this dataset is intended to be used for rephraser training.