{"id":"W7082252918","doi":"10.48448/04cz-9t45","title":"Memorization Inheritance in Sequence-Level Knowledge Distillation for Neural Machine Translation","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":"McGill University","funders":"","keywords":"Memorization; Counterfactual thinking; Machine translation; Quality (philosophy); Distillation; Inheritance (genetic algorithm); Task (project management)","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":[],"consensus_categories":[],"category_scores_codex":[0.0005742885,0.0002384795,0.0002431799,0.0004715663,0.0001586113,0.0001322057,0.001061532,0.0001919099,0.00004168771],"category_scores_gemma":[0.0002696699,0.0002324133,0.00004447569,0.001565257,0.0002185823,0.0003414638,0.0001291523,0.0001755884,0.000007732977],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001242582,"about_ca_system_score_gemma":0.0005141294,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009604424,"about_ca_topic_score_gemma":0.0006939946,"domain_scores_codex":[0.9982144,0.00003793378,0.0003339077,0.0007723109,0.0002737148,0.0003677277],"domain_scores_gemma":[0.9989758,0.0001065062,0.0002002209,0.0004552845,0.0001992962,0.00006296033],"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":[0.00001883803,0.0002208042,0.001009459,0.0007292774,0.00001970676,0.000008929005,0.0008983071,0.008945732,0.004386091,0.08836083,0.01140067,0.8840014],"study_design_scores_gemma":[0.0003458071,0.0000283696,0.0002598753,0.0001700633,0.000005876155,0.000002992535,0.00001328504,0.9270106,0.0003061555,0.005782984,0.06579514,0.0002788865],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0000158595,0.0007653108,0.7740898,0.001903939,0.0008232478,0.0005824834,0.0000642262,0.0002228168,0.2215323],"genre_scores_gemma":[0.5081878,0.000094555,0.1889621,0.0002333076,0.0004394248,0.0001233539,0.0003914628,0.00005153754,0.3015164],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9180648,"threshold_uncertainty_score":0.9477539,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07181544435982,"score_gpt":0.3166207451935005,"score_spread":0.2448053008336805,"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."}}