{"id":"W6967178806","doi":"10.48448/tkck-3g56","title":"Found in the Middle: Permutation Self-Consistency Improves Listwise Ranking in Large Language Models","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Permutation (music); Ranking (information retrieval); Robustness (evolution); Inference; Random permutation; Sorting; Resampling; Aggregate (composite)","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":[0.008824574,0.001628232,0.00191914,0.001370059,0.001458844,0.003534202,0.002935105,0.001715552,0.009993926],"category_scores_gemma":[0.0430687,0.0009966429,0.001811306,0.001686829,0.001806454,0.006145199,0.003600599,0.004308669,0.006862393],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001514435,"about_ca_system_score_gemma":0.003416635,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007403133,"about_ca_topic_score_gemma":0.01481894,"domain_scores_codex":[0.9918776,0.004972461,0.0003413239,0.001567324,0.0008725471,0.0003688461],"domain_scores_gemma":[0.974666,0.01292956,0.0007832238,0.009343487,0.00166906,0.0006087341],"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":[0.00212426,0.0006863719,0.02021038,0.0006589667,0.0009901624,0.0004296836,0.0009655323,0.2349615,0.008522085,0.1052913,0.1022047,0.5229549],"study_design_scores_gemma":[0.0002160425,0.0001990254,0.001121426,0.00005942626,0.00008991617,0.0001109307,0.0001007201,0.8059689,0.004922181,0.1791517,0.007991235,0.0000683228],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04928463,0.001326533,0.9142452,0.002237026,0.000400862,0.0001781466,0.001959899,0.02568365,0.004684091],"genre_scores_gemma":[0.5154137,0.0004401504,0.4556635,0.002343767,0.0006647177,0.0003856711,0.009003738,0.007061644,0.009023109],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009993926,"threshold_uncertainty_score":0.04666936,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03289868808050468,"score_gpt":0.304755673298126,"score_spread":0.2718569852176214,"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."}}