{"id":"W4416551090","doi":"10.48550/arxiv.2504.12083","title":"Self-alignment of Large Video Language Models with Refined Regularized Preference Optimization","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Mitacs","keywords":"Spurious relationship; Regularization (linguistics); Preference; Set (abstract data type); Simple (philosophy); Optimization problem","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0004072421,0.0003339241,0.0004497877,0.0001958209,0.0001245066,0.00007283843,0.001629919,0.0002416326,0.00001784434],"category_scores_gemma":[0.0000610915,0.0003015032,0.00009665088,0.0004520587,0.0000339125,0.0002044986,0.001714842,0.0005580244,0.00001308255],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001302289,"about_ca_system_score_gemma":0.0003332731,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004662362,"about_ca_topic_score_gemma":0.00002324154,"domain_scores_codex":[0.997636,0.0001968672,0.0004800771,0.0009465506,0.0004253688,0.0003151661],"domain_scores_gemma":[0.9968646,0.0001140808,0.0005154716,0.002162243,0.000253165,0.00009049052],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00003950817,0.0004362886,0.01394662,0.0004970313,0.0001985242,0.000005407869,0.003401201,0.9443967,0.0005873619,0.03562594,0.00008454901,0.0007808273],"study_design_scores_gemma":[0.0007200027,0.00004790424,0.01059324,0.0002718149,0.00006338276,0.00000152912,0.0000231412,0.9855036,0.001461607,0.0008843021,0.000111852,0.0003176766],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1411732,0.00009868811,0.8541145,0.0009730953,0.00009451741,0.0007549766,0.00003936489,0.0005146037,0.002237091],"genre_scores_gemma":[0.6019744,0.00002987063,0.3971023,0.0001264643,0.00002250057,0.0002144049,0.00008421659,0.00001519388,0.0004306472],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.4608012,"threshold_uncertainty_score":0.9999437,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02632753695901735,"score_gpt":0.2761713555977348,"score_spread":0.2498438186387175,"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."}}