{"id":"W4401043738","doi":"10.18653/v1/2024.naacl-short.16","title":"CELI: Simple yet Effective Approach to Enhance Out-of-Domain Generalization of Cross-Encoders.","year":2024,"lang":"en","type":"article","venue":"","topic":"Model Reduction and Neural Networks","field":"Physics and Astronomy","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Generalization; Simple (philosophy); Encoder; Computer science; Domain (mathematical analysis); Algorithm; Theoretical computer science; Mathematics","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":[0.00008753284,0.00009348054,0.0001432294,0.00004582988,0.00003020145,0.00003178395,0.00007187371,0.00002770725,0.0002569597],"category_scores_gemma":[0.000002249735,0.00007786485,0.00008134958,0.0001994028,0.00003507285,0.00009538812,0.00003041933,0.00006741864,0.00001711641],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000009659428,"about_ca_system_score_gemma":0.00001815843,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006762035,"about_ca_topic_score_gemma":0.000001441728,"domain_scores_codex":[0.9993361,0.00003445022,0.0001870936,0.0002160366,0.0001038179,0.0001224704],"domain_scores_gemma":[0.9996912,0.00003330357,0.0000382272,0.0001312384,0.0000576568,0.00004832758],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0001222678,0.0005716266,0.004726002,0.0003794563,0.0003273918,8.139983e-7,0.00483814,0.1851933,0.104465,0.5115472,0.06154795,0.1262808],"study_design_scores_gemma":[0.0004476119,0.0002328998,0.0007141536,0.0001045926,0.00005116767,7.217213e-7,0.0006589376,0.2678074,0.6672536,0.03577892,0.02638256,0.0005675323],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2338416,0.00004401726,0.7298462,0.00004047791,0.0003308964,0.0002782937,0.00001871568,0.00002821423,0.03557158],"genre_scores_gemma":[0.9946153,0.00000194668,0.003008821,0.00003292646,0.0001856425,0.00004084979,0.00004300996,0.00001269786,0.002058866],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7607737,"threshold_uncertainty_score":0.3175237,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01472647320876421,"score_gpt":0.3196598497462922,"score_spread":0.3049333765375281,"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."}}