{"id":"W3200285169","doi":"10.18653/v1/2021.emnlp-main.779","title":"PICARD: Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Models","year":2021,"lang":"en","type":"preprint","venue":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; University of Toronto","funders":"","keywords":"Computer science; Decoding methods; Parsing; Language model; Natural language processing; Artificial intelligence; Code (set theory); Programming language; Compiler; Algorithm; Set (abstract data type)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001926959,0.003057839,0.001384508,0.001231691,0.0007987178,0.002664651,0.00312672,0.002208459,0.01455098],"category_scores_gemma":[0.01530036,0.001935573,0.0020474,0.001049302,0.001256367,0.003214001,0.003295782,0.004347242,0.008778794],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001160997,"about_ca_system_score_gemma":0.003515273,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009828807,"about_ca_topic_score_gemma":0.02609289,"domain_scores_codex":[0.9982396,0.0005936691,0.0001151244,0.0005737827,0.0003400421,0.0001377341],"domain_scores_gemma":[0.9947523,0.003542662,0.0001432899,0.0009580164,0.0004985082,0.0001052411],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006691882,0.0002255013,0.002398109,0.0008250527,0.0004597921,0.0006857183,0.000682593,0.2430716,0.02887361,0.02716466,0.07495188,0.6199923],"study_design_scores_gemma":[0.00005583789,0.00004607507,0.0001768212,0.00003722253,0.00004491465,0.0001352851,0.00005608327,0.9584664,0.01325303,0.01973841,0.007943274,0.00004669802],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.007871318,0.0003333483,0.899148,0.0002669955,0.0001430389,0.000136701,0.00114753,0.08873478,0.002218243],"genre_scores_gemma":[0.1428921,0.0003146261,0.827105,0.0006312913,0.0001262331,0.000502458,0.00692154,0.0158948,0.005611944],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01455098,"threshold_uncertainty_score":0.04867786,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07064381815737095,"score_gpt":0.4144700036335565,"score_spread":0.3438261854761855,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}