{"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/what-do-recurrent-neural-network-grammars","title":"What Do Recurrent Neural Network Grammars Learn About Syntax?","arxiv_id":"1611.05774","date":"2016-11-17","proceeding":"EACL 2017 4","authors":["Adhiguna Kuncoro","Miguel Ballesteros","Lingpeng Kong","Chris Dyer","Graham Neubig","Noah A. Smith"],"abstract":"Recurrent neural network grammars (RNNG) are a recently proposed\nprobabilistic generative modeling family for natural language. They show\nstate-of-the-art language modeling and parsing performance. We investigate what\ninformation they learn, from a linguistic perspective, through various\nablations to the model and the data, and by augmenting the model with an\nattention mechanism (GA-RNNG) to enable closer inspection. We find that\nexplicit modeling of composition is crucial for achieving the best performance.\nThrough the attention mechanism, we find that headedness plays a central role\nin phrasal representation (with the model's latent attention largely agreeing\nwith predictions made by hand-crafted head rules, albeit with some important\ndifferences). By training grammars without nonterminal labels, we find that\nphrasal representations depend minimally on nonterminals, providing support for\nthe endocentricity hypothesis.","url_abs":"http://arxiv.org/abs/1611.05774v2","url_pdf":"http://arxiv.org/pdf/1611.05774v2.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":"what-do-recurrent-neural-network-grammars","repo_url":"https://github.com/clab/rnng","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"constituency-parsing","task_name":"Constituency Parsing"},{"task_slug":"dependency-parsing","task_name":"Dependency Parsing"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/constituency-parsing-on-penn-treebank","task":"Constituency Parsing","dataset":"Penn Treebank","model":"Stack-only RNNG","rank_in_archive_order":22,"of":27,"metrics":{"F1 score":"93.6"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.05774","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}