{"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/do-latent-tree-learning-models-identify","title":"Do latent tree learning models identify meaningful structure in sentences?","arxiv_id":"1709.01121","date":"2017-09-04","proceeding":"TACL 2018 1","authors":["Adina Williams","Andrew Drozdov","Samuel R. Bowman"],"abstract":"Recent work on the problem of latent tree learning has made it possible to\ntrain neural networks that learn to both parse a sentence and use the resulting\nparse to interpret the sentence, all without exposure to ground-truth parse\ntrees at training time. Surprisingly, these models often perform better at\nsentence understanding tasks than models that use parse trees from conventional\nparsers. This paper aims to investigate what these latent tree learning models\nlearn. We replicate two such models in a shared codebase and find that (i) only\none of these models outperforms conventional tree-structured models on sentence\nclassification, (ii) its parsing strategies are not especially consistent\nacross random restarts, (iii) the parses it produces tend to be shallower than\nstandard Penn Treebank (PTB) parses, and (iv) they do not resemble those of PTB\nor any other semantic or syntactic formalism that the authors are aware of.","url_abs":"http://arxiv.org/abs/1709.01121v2","url_pdf":"http://arxiv.org/pdf/1709.01121v2.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":"do-latent-tree-learning-models-identify","repo_url":"https://github.com/NYU-MLL/spinn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"sentence-classification","task_name":"Sentence Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}