{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/soft-layer-specific-multi-task-summarization","title":"Soft Layer-Specific Multi-Task Summarization with Entailment and Question Generation","arxiv_id":"1805.11004","date":"2018-05-28","proceeding":"ACL 2018 7","authors":["Han Guo","Ramakanth Pasunuru","Mohit Bansal"],"abstract":"An accurate abstractive summary of a document should contain all its salient\ninformation and should be logically entailed by the input document. We improve\nthese important aspects of abstractive summarization via multi-task learning\nwith the auxiliary tasks of question generation and entailment generation,\nwhere the former teaches the summarization model how to look for salient\nquestioning-worthy details, and the latter teaches the model how to rewrite a\nsummary which is a directed-logical subset of the input document. We also\npropose novel multi-task architectures with high-level (semantic)\nlayer-specific sharing across multiple encoder and decoder layers of the three\ntasks, as well as soft-sharing mechanisms (and show performance ablations and\nanalysis examples of each contribution). Overall, we achieve statistically\nsignificant improvements over the state-of-the-art on both the CNN/DailyMail\nand Gigaword datasets, as well as on the DUC-2002 transfer setup. We also\npresent several quantitative and qualitative analysis studies of our model's\nlearned saliency and entailment skills.","url_abs":"http://arxiv.org/abs/1805.11004v1","url_pdf":"http://arxiv.org/pdf/1805.11004v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"abstractive-text-summarization","task_name":"Abstractive Text Summarization"},{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"},{"task_slug":"question-generation","task_name":"Question Generation"},{"task_slug":"question-generation","task_name":"Question-Generation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/abstractive-text-summarization-on-cnn-daily","task":"Abstractive Text Summarization","dataset":"CNN / Daily Mail","model":"Pointer + Coverage + EntailmentGen + QuestionGen","rank_in_archive_order":46,"of":53,"metrics":{"ROUGE-1":"39.81","ROUGE-2":"17.64","ROUGE-L":"36.54"},"uses_additional_data":false},{"leaderboard":"/sota/text-summarization-on-gigaword","task":"Text Summarization","dataset":"GigaWord","model":"Pointer + Coverage + EntailmentGen + QuestionGen","rank_in_archive_order":35,"of":41,"metrics":{"ROUGE-1":"35.98","ROUGE-2":"17.76","ROUGE-L":"33.63"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.11004","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}