{"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/ruminating-reader-reasoning-with-gated-multi","title":"Ruminating Reader: Reasoning with Gated Multi-Hop Attention","arxiv_id":"1704.07415","date":"2017-04-24","proceeding":"WS 2018 7","authors":["Yichen Gong","Samuel R. Bowman"],"abstract":"To answer the question in machine comprehension (MC) task, the models need to\nestablish the interaction between the question and the context. To tackle the\nproblem that the single-pass model cannot reflect on and correct its answer, we\npresent Ruminating Reader. Ruminating Reader adds a second pass of attention\nand a novel information fusion component to the Bi-Directional Attention Flow\nmodel (BiDAF). We propose novel layer structures that construct an query-aware\ncontext vector representation and fuse encoding representation with\nintermediate representation on top of BiDAF model. We show that a multi-hop\nattention mechanism can be applied to a bi-directional attention structure. In\nexperiments on SQuAD, we find that the Reader outperforms the BiDAF baseline by\na substantial margin, and matches or surpasses the performance of all other\npublished systems.","url_abs":"http://arxiv.org/abs/1704.07415v1","url_pdf":"http://arxiv.org/pdf/1704.07415v1.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":"question-answering","task_name":"Question Answering"},{"task_slug":"reading-comprehension","task_name":"Reading Comprehension"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/question-answering-on-squad11","task":"Question Answering","dataset":"SQuAD1.1","model":"Ruminating Reader (single model)","rank_in_archive_order":159,"of":213,"metrics":{"EM":"70.639","F1":"79.456"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-squad11-dev","task":"Question Answering","dataset":"SQuAD1.1 dev","model":"Ruminating Reader","rank_in_archive_order":36,"of":55,"metrics":{"EM":"70.6","F1":"79.5"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}