{"id":"W4400406053","doi":"10.2139/ssrn.4856254","title":"Filtered not Mixed: Stochastic Filtering-Based Online Gating for Mixture of Large Language Models","year":2024,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Topic Modeling","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"Dalhousie University; Vector Institute; McMaster University; University of Toronto","funders":"","keywords":"Computer science; Gating; Language model; Artificial intelligence; Natural language processing; Psychology; Neuroscience","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":[0.004981416,0.001582171,0.002371494,0.001403738,0.0009517496,0.002331505,0.003341046,0.002647808,0.005977649],"category_scores_gemma":[0.01763398,0.001691391,0.001669361,0.001952078,0.001037979,0.004418087,0.004188984,0.003447267,0.002177536],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000974547,"about_ca_system_score_gemma":0.00264724,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005479844,"about_ca_topic_score_gemma":0.009750675,"domain_scores_codex":[0.9978883,0.0009040717,0.0001143703,0.0004499151,0.0003473587,0.0002960931],"domain_scores_gemma":[0.9913388,0.006694763,0.0003072736,0.0007923039,0.0004803546,0.0003864365],"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":[0.001940837,0.0006295802,0.002779899,0.0004095401,0.0003085339,0.0004001822,0.000473587,0.3615335,0.01796162,0.09927269,0.01408112,0.5002089],"study_design_scores_gemma":[0.00003089864,0.00003712132,0.0001619654,0.000009902205,0.00002103432,0.00003254742,0.000008589151,0.9769709,0.001103424,0.02084264,0.0007653058,0.00001570797],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006615459,0.0001846707,0.9911643,0.0001727069,0.00007493224,0.00004208195,0.000166933,0.00116152,0.0004173665],"genre_scores_gemma":[0.4147515,0.0007124802,0.5703058,0.0008899457,0.0006520793,0.0005781192,0.003133364,0.001326673,0.00764989],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005977649,"threshold_uncertainty_score":0.02634454,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02135863256316978,"score_gpt":0.2869592703647592,"score_spread":0.2656006378015894,"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."}}