{"url":"/sota/constituency-grammar-induction-on-ptb","task":{"name":"Constituency Grammar Induction","url":"/task/constituency-grammar-induction","note":null},"dataset":{"name":"PTB Diagnostic ECG Database","url":"/dataset/ptb"},"category":"Natural Language Processing","categories":["Natural Language Processing"],"category_note":null,"description":"Inducing a constituency-based phrase structure grammar.","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Mean F1 (WSJ)","Max F1 (WSJ)","Mean F1 (WSJ10)","Max F1 (WSJ10)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Mean F1 (WSJ)":"higher","Max F1 (WSJ)":"higher","Mean F1 (WSJ10)":"higher","Max F1 (WSJ10)":"higher"}},"counts":{"rows":24,"rows_with_code":23,"rows_with_paper_page":24,"rows_dated":23,"rows_using_additional_data":1},"rows":[{"rank_in_archive_order":1,"model":"Ensemble (Generative MBR)","metrics":{"Max F1 (WSJ)":"71.9","Mean F1 (WSJ)":"70.4"},"uses_additional_data":false,"paper_date":"2023-10-03","paper":"/paper/ensemble-distillation-for-unsupervised","paper_url":"https://arxiv.org/abs/2310.01717v2","paper_title":"Ensemble Distillation for Unsupervised Constituency Parsing","code":"https://github.com/manga-uofa/ed4ucp","n_code_links":1,"syntology":{"n_ran":16,"n_unverified":6,"n_samples":22,"n_pointer_only_licence":22}},{"rank_in_archive_order":2,"model":"Parse-Focused (NT=4500)","metrics":{"Max F1 (WSJ)":"70.3","Mean F1 (WSJ)":"69.6"},"uses_additional_data":false,"paper_date":"2024-07-23","paper":"/paper/structural-optimization-ambiguity-and","paper_url":"https://arxiv.org/abs/2407.16181v1","paper_title":"Structural Optimization Ambiguity and Simplicity Bias in Unsupervised Neural Grammar Induction","code":"https://github.com/GIST-IRR/Parse-Focusing","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"Parse-Focused (NT=30)","metrics":{"Max F1 (WSJ)":"68.4","Mean F1 (WSJ)":"67.4"},"uses_additional_data":false,"paper_date":"2024-07-23","paper":"/paper/structural-optimization-ambiguity-and","paper_url":"https://arxiv.org/abs/2407.16181v1","paper_title":"Structural Optimization Ambiguity and Simplicity Bias in Unsupervised Neural Grammar Induction","code":"https://github.com/GIST-IRR/Parse-Focusing","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"SemInfo-SCPCFG (1024NT)","metrics":{"Mean F1 (WSJ)":"66.92"},"uses_additional_data":false,"paper_date":"2024-10-03","paper":"/paper/improving-unsupervised-constituency-parsing","paper_url":"https://arxiv.org/abs/2410.02558v2","paper_title":"Improving Unsupervised Constituency Parsing via Maximizing Semantic Information","code":"https://github.com/junjiechen-chris/improving-unsupervised-constituency-parsing-via-maximizing-semantic-information","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":5,"model":"Ensemble (Selective MBR)","metrics":{"Mean F1 (WSJ)":"66.2"},"uses_additional_data":false,"paper_date":"2023-10-03","paper":"/paper/ensemble-distillation-for-unsupervised","paper_url":"https://arxiv.org/abs/2310.01717v2","paper_title":"Ensemble Distillation for Unsupervised Constituency Parsing","code":"https://github.com/manga-uofa/ed4ucp","n_code_links":1,"syntology":{"n_ran":16,"n_unverified":6,"n_samples":22,"n_pointer_only_licence":22}},{"rank_in_archive_order":6,"model":"ReCAT","metrics":{"Mean F1 (WSJ)":"65.0"},"uses_additional_data":false,"paper_date":"2023-09-28","paper":"/paper/augmenting-transformers-with-recursively","paper_url":"https://arxiv.org/abs/2309.16319v2","paper_title":"Augmenting Transformers with Recursively Composed Multi-grained Representations","code":"https://github.com/ant-research/structuredlm_rtdt","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"DP in rank space","metrics":{"Mean F1 (WSJ)":"64.1"},"uses_additional_data":false,"paper_date":"2022-05-01","paper":"/paper/dynamic-programming-in-rank-space-scaling-1","paper_url":"https://arxiv.org/abs/2205.00484v1","paper_title":"Dynamic Programming in Rank Space: Scaling Structured Inference with Low-Rank HMMs and PCFGs","code":"https://github.com/sustcsonglin/TN-PCFG","n_code_links":2,"syntology":null},{"rank_in_archive_order":8,"model":"inside-outside co-training + weak supervision","metrics":{"Max F1 (WSJ)":"66.8","Mean F1 (WSJ)":"63.1","Mean F1 (WSJ10)":"74.2"},"uses_additional_data":true,"paper_date":"2021-10-05","paper":"/paper/co-training-an-unsupervised-constituency","paper_url":"https://arxiv.org/abs/2110.02283v2","paper_title":"Co-training an Unsupervised Constituency Parser with Weak Supervision","code":"https://github.com/Nickil21/weakly-supervised-parsing","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":1,"n_samples":3,"n_pointer_only_licence":1}},{"rank_in_archive_order":9,"model":"Hashing (Parserker 2)","metrics":{"Max F1 (WSJ)":"64.1","Mean F1 (WSJ)":"62.4"},"uses_additional_data":false,"paper_date":"2024-10-05","paper":"/paper/on-eliciting-syntax-from-language-models-via","paper_url":"https://arxiv.org/abs/2410.04074v1","paper_title":"On Eliciting Syntax from Language Models via Hashing","code":"https://github.com/speedcell4/parserker","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"NBL-PCFG","metrics":{"Mean F1 (WSJ)":"60.4"},"uses_additional_data":false,"paper_date":"2021-05-31","paper":"/paper/neural-bi-lexicalized-pcfg-induction","paper_url":"https://arxiv.org/abs/2105.15021v1","paper_title":"Neural Bi-Lexicalized PCFG Induction","code":"https://github.com/sustcsonglin/TN-PCFG","n_code_links":1,"syntology":null},{"rank_in_archive_order":11,"model":"TN-PCFG (p=500)","metrics":{"Max F1 (WSJ)":"61.4","Mean F1 (WSJ)":"57.7"},"uses_additional_data":false,"paper_date":"2021-04-28","paper":"/paper/pcfgs-can-do-better-inducing-probabilistic","paper_url":"https://arxiv.org/abs/2104.13727v1","paper_title":"PCFGs Can Do Better: Inducing Probabilistic Context-Free Grammars with Many Symbols","code":"https://github.com/sustcsonglin/TN-PCFG","n_code_links":1,"syntology":null},{"rank_in_archive_order":12,"model":"S-DIORA","metrics":{"Max F1 (WSJ)":"63.96","Max F1 (WSJ10)":"71.8","Mean F1 (WSJ)":"57.6"},"uses_additional_data":false,"paper_date":null,"paper":"/paper/unsupervised-parsing-with-s-diora-single-tree","paper_url":"https://aclanthology.org/2020.emnlp-main.392","paper_title":"Unsupervised Parsing with S-DIORA: Single Tree Encoding for Deep Inside-Outside Recursive Autoencoders","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":13,"model":"Fast-R2D2","metrics":{"Mean F1 (WSJ)":"57.2"},"uses_additional_data":false,"paper_date":"2022-03-01","paper":"/paper/fast-r2d2-a-pretrained-recursive-neural","paper_url":"https://arxiv.org/abs/2203.00281v3","paper_title":"Fast-R2D2: A Pretrained Recursive Neural Network based on Pruned CKY for Grammar Induction and Text Representation","code":"https://github.com/alipay/StructuredLM_RTDT","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":14,"model":"DIORA (+PP)","metrics":{"Max F1 (WSJ)":"56.2","Max F1 (WSJ10)":"60.55","Mean F1 (WSJ)":"55.7"},"uses_additional_data":false,"paper_date":"2019-06-01","paper":"/paper/unsupervised-latent-tree-induction-with-deep-1","paper_url":"https://aclanthology.org/N19-1116","paper_title":"Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Auto-Encoders","code":"https://github.com/iesl/diora","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"Compound PCFG","metrics":{"Max F1 (WSJ)":"60.1","Max F1 (WSJ10)":"68.8","Mean F1 (WSJ)":"55.2"},"uses_additional_data":false,"paper_date":"2019-06-24","paper":"/paper/compound-probabilistic-context-free-grammars","paper_url":"https://arxiv.org/abs/1906.10225v9","paper_title":"Compound Probabilistic Context-Free Grammars for Grammar Induction","code":"https://github.com/harvardnlp/compound-pcfg","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":16,"model":"GPST(left to right parsing)","metrics":{"Mean F1 (WSJ)":"55.2"},"uses_additional_data":false,"paper_date":"2024-03-13","paper":"/paper/generative-pretrained-structured-transformers","paper_url":"https://arxiv.org/abs/2403.08293v3","paper_title":"Generative Pretrained Structured Transformers: Unsupervised Syntactic Language Models at Scale","code":"https://github.com/ant-research/structuredlm_rtdt","n_code_links":2,"syntology":{"n_ran":4,"n_unverified":3,"n_samples":7,"n_pointer_only_licence":0}},{"rank_in_archive_order":17,"model":"Neural PCFG","metrics":{"Max F1 (WSJ)":"52.6","Mean F1 (WSJ)":"50.8"},"uses_additional_data":false,"paper_date":"2019-06-24","paper":"/paper/compound-probabilistic-context-free-grammars","paper_url":"https://arxiv.org/abs/1906.10225v9","paper_title":"Compound Probabilistic Context-Free Grammars for Grammar Induction","code":"https://github.com/harvardnlp/compound-pcfg","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":18,"model":"DIORA","metrics":{"Max F1 (WSJ)":"49.6","Max F1 (WSJ10)":"68.5","Mean F1 (WSJ)":"48.9","Mean F1 (WSJ10)":"67.7"},"uses_additional_data":false,"paper_date":"2019-06-01","paper":"/paper/unsupervised-latent-tree-induction-with-deep-1","paper_url":"https://aclanthology.org/N19-1116","paper_title":"Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Auto-Encoders","code":"https://github.com/iesl/diora","n_code_links":1,"syntology":null},{"rank_in_archive_order":19,"model":"ON-LSTM (tuned)","metrics":{"Max F1 (WSJ)":"50.0","Mean F1 (WSJ)":"48.1"},"uses_additional_data":false,"paper_date":"2018-10-22","paper":"/paper/ordered-neurons-integrating-tree-structures","paper_url":"https://arxiv.org/abs/1810.09536v6","paper_title":"Ordered Neurons: Integrating Tree Structures into Recurrent Neural Networks","code":"https://github.com/yikangshen/Ordered-Neurons","n_code_links":7,"syntology":{"n_ran":3,"n_unverified":9,"n_samples":12,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"DMV + invertible projector","metrics":{"Mean F1 (WSJ)":"47.9","Mean F1 (WSJ10)":"60.2"},"uses_additional_data":false,"paper_date":"2018-08-28","paper":"/paper/unsupervised-learning-of-syntactic-structure","paper_url":"http://arxiv.org/abs/1808.09111v1","paper_title":"Unsupervised Learning of Syntactic Structure with Invertible Neural Projections","code":"https://github.com/jxhe/struct-learning-with-flow","n_code_links":1,"syntology":null},{"rank_in_archive_order":21,"model":"ON-LSTM","metrics":{"Max F1 (WSJ)":"49.4","Max F1 (WSJ10)":"66.8","Mean F1 (WSJ)":"47.7","Mean F1 (WSJ10)":"65.1"},"uses_additional_data":false,"paper_date":"2018-10-22","paper":"/paper/ordered-neurons-integrating-tree-structures","paper_url":"https://arxiv.org/abs/1810.09536v6","paper_title":"Ordered Neurons: Integrating Tree Structures into Recurrent Neural Networks","code":"https://github.com/yikangshen/Ordered-Neurons","n_code_links":7,"syntology":{"n_ran":3,"n_unverified":9,"n_samples":12,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"PRPN (tuned)","metrics":{"Max F1 (WSJ)":"47.9","Mean F1 (WSJ)":"47.3"},"uses_additional_data":false,"paper_date":"2017-11-02","paper":"/paper/neural-language-modeling-by-jointly-learning","paper_url":"http://arxiv.org/abs/1711.02013v2","paper_title":"Neural Language Modeling by Jointly Learning Syntax and Lexicon","code":"https://github.com/nyu-mll/PRPN-Analysis","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":5,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":23,"model":"URNNG","metrics":{"Max F1 (WSJ)":"52.4"},"uses_additional_data":false,"paper_date":"2019-04-07","paper":"/paper/unsupervised-recurrent-neural-network","paper_url":"https://arxiv.org/abs/1904.03746v6","paper_title":"Unsupervised Recurrent Neural Network Grammars","code":"https://github.com/harvardnlp/urnng","n_code_links":1,"syntology":null},{"rank_in_archive_order":24,"model":"PRPN","metrics":{"Max F1 (WSJ)":"38.1"},"uses_additional_data":false,"paper_date":"2017-11-02","paper":"/paper/neural-language-modeling-by-jointly-learning","paper_url":"http://arxiv.org/abs/1711.02013v2","paper_title":"Neural Language Modeling by Jointly Learning Syntax and Lexicon","code":"https://github.com/nyu-mll/PRPN-Analysis","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":5,"n_samples":5,"n_pointer_only_licence":0}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":13,"rows_with_any_sample_ran":10,"distinct_papers_with_graph_line":9,"distinct_papers_with_any_sample_ran":7,"samples_over_distinct_papers":{"n_ran":30,"n_unverified":25,"n_samples":55,"n_pointer_only_licence":25,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":50,"n_unverified":45,"n_samples":95,"n_pointer_only_licence":48,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}