The dataset is generated from
https://huggingface.co/datasets/openai/summarize_from_feedback.
Please refer to
https://github.com/liyuan24/dgx_spark_summary_from_human_feedback/tree/main?tab=readme-ov-file#download-the-comparison-dataset about how to download the dataset.
This is a comparison dataset used for training a reward model. Each example contains a query (post) and two responses (chosen and rejected) where… See the full description on the dataset page:
https://huggingface.co/datasets/seangogo/processed_tldr_comparison_dataset_20251102_065554.