{"id":"W4389519395","doi":"10.18653/v1/2023.emnlp-main.195","title":"Pushdown Layers: Encoding Recursive Structure in Transformer Language Models","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Center for Evolutionary and Theoretical Immunology; Canadian Institute for Advanced Research","keywords":"Computer science; Parsing; Transformer; Stack (abstract data type); Artificial intelligence; Algorithm; Theoretical computer science; Programming language","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009685779,0.001278334,0.0005295877,0.0007355437,0.0003770464,0.001318085,0.001563512,0.0008545882,0.005586636],"category_scores_gemma":[0.005071423,0.0006853823,0.001065259,0.0006096157,0.0006550262,0.003497622,0.001791778,0.002208162,0.002661827],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009338008,"about_ca_system_score_gemma":0.001031638,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00630812,"about_ca_topic_score_gemma":0.01603903,"domain_scores_codex":[0.9996412,0.00009472302,0.00002551114,0.0001304835,0.00006305162,0.00004502032],"domain_scores_gemma":[0.998708,0.0007864541,0.00005757422,0.0002324028,0.0001627428,0.00005270877],"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":[0.0005782736,0.0001521839,0.004299123,0.0003745272,0.0002090236,0.0004371874,0.0007434844,0.3342288,0.03008236,0.04028007,0.01826582,0.5703493],"study_design_scores_gemma":[0.00002341036,0.00003993189,0.0002378155,0.00001875499,0.00003470732,0.0000459285,0.00003430447,0.9622893,0.007392883,0.02695955,0.002904055,0.00001933171],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03856281,0.0004837677,0.9383884,0.0003319225,0.0001229291,0.0001131996,0.001536417,0.01712284,0.003337696],"genre_scores_gemma":[0.6863447,0.0005935683,0.2961908,0.0005373892,0.00008283792,0.0003587033,0.004380757,0.002667262,0.008843911],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00630812,"threshold_uncertainty_score":0.0186891,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02570324384620903,"score_gpt":0.2635919863223192,"score_spread":0.2378887424761102,"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."}}