{"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/synchronous-bidirectional-inference-for","title":"Synchronous Bidirectional Inference for Neural Sequence Generation","arxiv_id":"1902.08955","date":"2019-02-24","proceeding":null,"authors":["Jiajun Zhang","Long Zhou","Yang Zhao","Cheng-qing Zong"],"abstract":"In sequence to sequence generation tasks (e.g. machine translation and\nabstractive summarization), inference is generally performed in a left-to-right\nmanner to produce the result token by token. The neural approaches, such as\nLSTM and self-attention networks, are now able to make full use of all the\npredicted history hypotheses from left side during inference, but cannot\nmeanwhile access any future (right side) information and usually generate\nunbalanced outputs in which left parts are much more accurate than right ones.\nIn this work, we propose a synchronous bidirectional inference model to\ngenerate outputs using both left-to-right and right-to-left decoding\nsimultaneously and interactively. First, we introduce a novel beam search\nalgorithm that facilitates synchronous bidirectional decoding. Then, we present\nthe core approach which enables left-to-right and right-to-left decoding to\ninteract with each other, so as to utilize both the history and future\npredictions simultaneously during inference. We apply the proposed model to\nboth LSTM and self-attention networks. In addition, we propose two strategies\nfor parameter optimization. The extensive experiments on machine translation\nand abstractive summarization demonstrate that our synchronous bidirectional\ninference model can achieve remarkable improvements over the strong baselines.","url_abs":"http://arxiv.org/abs/1902.08955v1","url_pdf":"http://arxiv.org/pdf/1902.08955v1.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":[{"paper_slug":"synchronous-bidirectional-inference-for","repo_url":"https://github.com/ZNLP/sb-nmt","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"abstractive-text-summarization","task_name":"Abstractive Text Summarization"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"translation","task_name":"Translation"}],"methods":[{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.08955","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}