{"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/exploring-the-syntactic-abilities-of-rnns","title":"Exploring the Syntactic Abilities of RNNs with Multi-task Learning","arxiv_id":"1706.03542","date":"2017-06-12","proceeding":"CONLL 2017 8","authors":["Emile Enguehard","Yoav Goldberg","Tal Linzen"],"abstract":"Recent work has explored the syntactic abilities of RNNs using the\nsubject-verb agreement task, which diagnoses sensitivity to sentence structure.\nRNNs performed this task well in common cases, but faltered in complex\nsentences (Linzen et al., 2016). We test whether these errors are due to\ninherent limitations of the architecture or to the relatively indirect\nsupervision provided by most agreement dependencies in a corpus. We trained a\nsingle RNN to perform both the agreement task and an additional task, either\nCCG supertagging or language modeling. Multi-task training led to significantly\nlower error rates, in particular on complex sentences, suggesting that RNNs\nhave the ability to evolve more sophisticated syntactic representations than\nshown before. We also show that easily available agreement training data can\nimprove performance on other syntactic tasks, in particular when only a limited\namount of training data is available for those tasks. The multi-task paradigm\ncan also be leveraged to inject grammatical knowledge into language models.","url_abs":"http://arxiv.org/abs/1706.03542v1","url_pdf":"http://arxiv.org/pdf/1706.03542v1.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":"exploring-the-syntactic-abilities-of-rnns","repo_url":"https://github.com/emengd/multitask-agreement","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"ccg-supertagging","task_name":"CCG Supertagging"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"},{"task_slug":"sentence","task_name":"Sentence"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.03542","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}