{"id":"W4407244642","doi":"10.48550/arxiv.2502.03699","title":"LLM Alignment as Retriever Optimization: An Information Retrieval Perspective","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Office of Naval Research; Defense Advanced Research Projects Agency; Cisco Systems; National Science Foundation","keywords":"Perspective (graphical); Labrador Retriever; Information retrieval; Computer science; Medicine; Artificial intelligence","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0006751341,0.0003194278,0.0003250378,0.0004155293,0.0002804573,0.001124456,0.002331522,0.0003330411,0.0001312856],"category_scores_gemma":[0.0001678353,0.0002998156,0.0001369225,0.001082757,0.00007317754,0.01631842,0.001654739,0.0004355391,0.0006100581],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003637723,"about_ca_system_score_gemma":0.001020784,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001187428,"about_ca_topic_score_gemma":6.517328e-7,"domain_scores_codex":[0.9972982,0.0001319748,0.0008095233,0.0005448382,0.0008688213,0.00034668],"domain_scores_gemma":[0.9971368,0.0000459693,0.0005665418,0.001425628,0.000628008,0.0001970228],"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":[0.0002325242,0.0005172097,0.05003388,0.0006274793,0.0004296141,0.00003052826,0.1853084,0.438778,0.00004599882,0.276382,0.04273666,0.004877773],"study_design_scores_gemma":[0.00178985,0.0009019066,0.02452397,0.0006506044,0.0000376018,0.00005160136,0.01389551,0.8417668,0.006274234,0.007654083,0.09998368,0.002470203],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0783189,0.0001501675,0.7694813,0.006775249,0.006269117,0.00157752,0.00006714035,0.0009436746,0.1364169],"genre_scores_gemma":[0.9559013,0.0002385554,0.0228266,0.01209481,0.0006500015,0.00008074439,0.0004549693,0.00001827675,0.007734702],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8775824,"threshold_uncertainty_score":0.9999454,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02837275094583898,"score_gpt":0.2715936308975445,"score_spread":0.2432208799517055,"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."}}