{"id":"W4389523905","doi":"10.18653/v1/2023.emnlp-main.83","title":"How Does Generative Retrieval Scale to Millions of Passages?","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":23,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Generative grammar; Transformer; Information retrieval; Search engine indexing; Document retrieval; Artificial intelligence; Encoder; Generative model; Ranking (information retrieval); Task (project management); Natural language processing","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":[0.005725068,0.0009495816,0.001912697,0.001985125,0.001037065,0.004294151,0.002311803,0.002187183,0.004953386],"category_scores_gemma":[0.05261387,0.0009270436,0.001065672,0.002259317,0.002112512,0.01458731,0.002600882,0.002953768,0.004444962],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001146539,"about_ca_system_score_gemma":0.001051597,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004261485,"about_ca_topic_score_gemma":0.003892391,"domain_scores_codex":[0.9963416,0.001485269,0.0002233812,0.0008470521,0.0008357458,0.0002668724],"domain_scores_gemma":[0.9791266,0.01217877,0.0007178791,0.005718767,0.001814241,0.0004439121],"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":[0.0006570002,0.000566899,0.0107915,0.001485697,0.0003807905,0.000830912,0.002519825,0.1460179,0.03559697,0.06754228,0.0568368,0.6767735],"study_design_scores_gemma":[0.0001730689,0.0003950124,0.004175646,0.0001613396,0.0002289215,0.001387045,0.001275347,0.7431663,0.01948647,0.2006493,0.02875356,0.0001481275],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2004294,0.01110954,0.7324919,0.01475509,0.0009918824,0.000482706,0.001459305,0.01276462,0.02551567],"genre_scores_gemma":[0.777626,0.004684524,0.2027737,0.002145028,0.001002398,0.0003604979,0.002617021,0.001782977,0.007007901],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005725068,"threshold_uncertainty_score":0.03027743,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03023236006595731,"score_gpt":0.2627108354550887,"score_spread":0.2324784753891314,"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."}}