{"id":"W3152698349","doi":"10.18653/v1/2021.emnlp-main.230","title":"Masked Language Modeling and the Distributional Hypothesis: Order Word Matters Pre-training for Little","year":2021,"lang":"en","type":"article","venue":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","topic":"Topic Modeling","field":"Computer Science","cited_by":182,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Word (group theory); Natural language processing; Artificial intelligence; Word order; Language model; Downstream (manufacturing); Order (exchange); Parametric statistics; Linguistics; Mathematics","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001870851,0.0002261407,0.0004033016,0.00007884572,0.0002148773,0.0003685529,0.0009118242,0.0001115139,0.000007884013],"category_scores_gemma":[0.004471921,0.0001377908,0.0001166379,0.0007115842,0.0001589739,0.0003659466,0.0004990299,0.0005436991,3.208969e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006224908,"about_ca_system_score_gemma":0.0002046844,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001425403,"about_ca_topic_score_gemma":0.000003758969,"domain_scores_codex":[0.997997,0.0001289172,0.0004594805,0.0005843429,0.0004390718,0.0003911924],"domain_scores_gemma":[0.998121,0.0008075233,0.0002324658,0.000234176,0.0005454685,0.00005938307],"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":[0.0002065022,0.00007319539,0.0002971615,0.0004367492,0.00004784437,0.000004518339,0.03589173,0.0003752779,0.03054211,0.02637488,0.00003646452,0.9057136],"study_design_scores_gemma":[0.0008619824,0.0000123253,0.0001628048,0.0005348476,0.00002648223,0.00002456478,0.007942137,0.9533445,0.01130675,0.02552713,0.00003071303,0.0002257317],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1890912,0.004977314,0.7662941,0.03801954,0.0003275738,0.0005947074,0.00001068046,0.00007990595,0.0006049736],"genre_scores_gemma":[0.5688958,0.000008683237,0.4300471,0.0007740436,0.00006291036,0.00004386071,0.000001225209,0.000009917489,0.0001564631],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9529693,"threshold_uncertainty_score":0.5618944,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07636691541448742,"score_gpt":0.3862203416242639,"score_spread":0.3098534262097765,"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."}}