A package to generate summaries of long-form text and evaluate the coherence of these summaries. Official package for our ICLR 2024 paper, "BooookScore: A systematic exploration of book-length summarization in the era of LLMs".
I understand that the books cannot be released due to copyright reasons, but is there anyway that I can get the file containing all the books? One of my work would benefit much from this dataset for evaluation purposes. I would only use this dataset for research purposes and won't make it public. Thank you!
Hi
I understand that the books cannot be released due to copyright reasons, but is there anyway that I can get the file containing all the books? One of my work would benefit much from this dataset for evaluation purposes. I would only use this dataset for research purposes and won't make it public. Thank you!