{"id":"W4407590589","doi":"10.48550/arxiv.2502.09589","title":"Logical forms complement probability in understanding language model (and human) performance","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Complement (music); Computer science; Logical conjunction; Natural language processing; Programming language; Chemistry","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0006517291,0.00030524,0.0003805758,0.0001995327,0.0001638509,0.0001459471,0.001237097,0.00024427,0.000006074038],"category_scores_gemma":[0.00004634302,0.0002524225,0.00006792637,0.00022913,0.0001049952,0.0003221743,0.00341022,0.0009225962,0.0000019496],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005590294,"about_ca_system_score_gemma":0.0001320314,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001032948,"about_ca_topic_score_gemma":0.0001666767,"domain_scores_codex":[0.9980149,0.00005819397,0.0004310157,0.0008350069,0.000268646,0.0003921842],"domain_scores_gemma":[0.9988557,0.00005449542,0.000164307,0.0008109615,0.00004886711,0.00006566898],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0000377906,0.0003732123,0.4578443,0.003227108,0.00005862645,0.0000736891,0.007979229,0.0030354,0.001681061,0.5177049,0.0001754459,0.007809205],"study_design_scores_gemma":[0.0005981776,0.0001159253,0.02795267,0.001392377,0.00002102642,0.000008045439,0.0001505177,0.268325,0.003222944,0.6972654,0.0000258011,0.0009220891],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7206988,0.0005914263,0.2764713,0.0004251695,0.00006835835,0.0005177834,0.000008043307,0.000425267,0.0007937801],"genre_scores_gemma":[0.9109717,0.00002853838,0.0885461,0.0002243666,0.00002026701,0.00007881065,0.00001675795,0.000008030962,0.0001054085],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4298917,"threshold_uncertainty_score":0.9999928,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1211045059254114,"score_gpt":0.337995929592267,"score_spread":0.2168914236668557,"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."}}