{"id":"W7077908793","doi":"10.48448/tmm4-en83","title":"Cross-lingual Transfer of Reward Models in Multilingual Alignment","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Transfer (computing); Representation (politics); Transfer of learning; Limiting; Multilingualism; Reinforcement learning; Language model","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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00101871,0.0002874889,0.0004090292,0.0005828544,0.00008457949,0.0001011354,0.002414078,0.000264397,0.0001724265],"category_scores_gemma":[0.0001826511,0.0002676058,0.00007907009,0.001104556,0.0008005581,0.0002069018,0.0004992831,0.0003000157,0.000009893436],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008433161,"about_ca_system_score_gemma":0.001168127,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005094882,"about_ca_topic_score_gemma":0.0001558603,"domain_scores_codex":[0.9972467,0.00003700522,0.0005163293,0.0009636435,0.0006905798,0.0005457269],"domain_scores_gemma":[0.9986148,0.00007410529,0.0001244071,0.0008800048,0.0002039119,0.0001027244],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00007704849,0.002273476,0.002569001,0.002284845,0.0002020155,0.0004380615,0.01014937,0.1824244,0.01837586,0.5114655,0.01728483,0.2524557],"study_design_scores_gemma":[0.003170891,0.0002851994,0.0002628965,0.002291395,0.00004077839,0.00003351877,0.0004525275,0.6593508,0.1120464,0.07269467,0.1471523,0.002218686],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.001325837,0.0003947747,0.200335,0.0007242103,0.0008295098,0.0004320098,0.00003453125,0.0002446997,0.7956794],"genre_scores_gemma":[0.6075015,0.00007758648,0.03618061,0.0002660176,0.0001361053,0.00001895451,0.00001402119,0.00002972433,0.3557755],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.6061757,"threshold_uncertainty_score":0.9999776,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02398400321094358,"score_gpt":0.2973894837165375,"score_spread":0.2734054805055939,"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."}}