{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001403128,0.0006767197,0.000430372,0.0002271269,0.0003281522,0.00081155,0.000588985,0.0004454012,0.005274985],"category_scores_gemma":[0.008373985,0.000210874,0.0002947173,0.0002968518,0.0004887983,0.001196498,0.001719957,0.001156503,0.001812205],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004412901,"about_ca_system_score_gemma":0.0008297493,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003255826,"about_ca_topic_score_gemma":0.003462158,"domain_scores_codex":[0.9990147,0.0003984873,0.00004137238,0.0003107281,0.0001328728,0.0001018234],"domain_scores_gemma":[0.9976164,0.001257749,0.0002121093,0.0005152347,0.0002825395,0.0001158479],"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":[0.000909216,0.0008363464,0.01324228,0.0004114284,0.0001983869,0.0003748682,0.001165433,0.2773609,0.07816397,0.009084689,0.004752938,0.6134995],"study_design_scores_gemma":[0.00004354939,0.0006926897,0.008371325,0.00004324722,0.00005400322,0.0001392693,0.0003177812,0.9341829,0.03784344,0.01281175,0.005434341,0.00006570165],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6331345,0.0007424452,0.3350011,0.000586789,0.0002790329,0.0001365742,0.0004051677,0.005750236,0.02396405],"genre_scores_gemma":[0.9693075,0.0001005754,0.0266114,0.00009766972,0.00001543024,0.00005778159,0.0002606832,0.0002403361,0.003308564],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.005274985,"threshold_uncertainty_score":0.01764655,"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."}}