{"id":"W7111684846","doi":"","title":"MetaAligner::Towards Generalizable Multi-Objective Alignment of Language Models","year":2024,"lang":"en","type":"article","venue":"Research Explorer (The University of Manchester)","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Open Text (Canada)","funders":"","keywords":"Inference; Preference; Language model; Focus (optics); Training set; Machine translation","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":[0.00302638,0.002685767,0.001547129,0.00119694,0.000656112,0.001853314,0.002374739,0.001728613,0.006490262],"category_scores_gemma":[0.009049831,0.001318093,0.002372055,0.001129955,0.0008940916,0.003754186,0.002651179,0.004732026,0.0037461],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001179446,"about_ca_system_score_gemma":0.002587583,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005776524,"about_ca_topic_score_gemma":0.00903454,"domain_scores_codex":[0.9975625,0.0009252743,0.0001413157,0.0008002764,0.0004204547,0.0001501123],"domain_scores_gemma":[0.9977998,0.001195538,0.00017186,0.0004187468,0.0003194573,0.00009454764],"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":[0.0002695665,0.0001941887,0.00173964,0.0004996799,0.0003013341,0.0002223632,0.0003947025,0.4482286,0.008697036,0.01859988,0.01371204,0.5071411],"study_design_scores_gemma":[0.00002648834,0.00004810556,0.0001620052,0.00002325978,0.00002257758,0.00003640029,0.00003935239,0.9763435,0.002658748,0.0171957,0.003423568,0.0000203598],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.005943498,0.0004112507,0.9805207,0.000194284,0.00008242571,0.00009421942,0.0003311254,0.01099028,0.001432143],"genre_scores_gemma":[0.1878209,0.0005584712,0.7970866,0.0008469911,0.0001238783,0.0005173637,0.003182392,0.004064445,0.005798905],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006490262,"threshold_uncertainty_score":0.02171206,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1223938224634734,"score_gpt":0.3180229600980785,"score_spread":0.195629137634605,"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."}}