{"id":"W7110016663","doi":"10.21428/594757db.13f92bbd","title":"Let’s CONFER: A Dataset for Evaluating Natural Language Inference Models on CONditional InFERence and Presupposition","year":2025,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Ottawa; Carleton University","funders":"","keywords":"Presupposition; Inference; Generalization; Task (project management); Sentence; Natural language","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":[0.002382319,0.002546002,0.0008482562,0.002390568,0.001564524,0.001871401,0.004095525,0.003531096,0.01052891],"category_scores_gemma":[0.01552741,0.0005157691,0.00179629,0.002119546,0.0009901843,0.003634869,0.002432112,0.003445596,0.007018273],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001962452,"about_ca_system_score_gemma":0.002352041,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02398653,"about_ca_topic_score_gemma":0.06208686,"domain_scores_codex":[0.9978697,0.000606296,0.0002061548,0.0006893993,0.0004983451,0.0001300996],"domain_scores_gemma":[0.9942821,0.002824751,0.0003159195,0.001341746,0.00082601,0.0004095355],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001638225,0.001566895,0.01339184,0.003394244,0.0005662522,0.0008423409,0.0007319194,0.02114833,0.008017856,0.009416594,0.8209827,0.1183028],"study_design_scores_gemma":[0.001841541,0.001081072,0.04291297,0.000616966,0.0004853968,0.001867583,0.00164761,0.3300138,0.02838642,0.03047474,0.5602354,0.0004365258],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.1011323,0.004231276,0.0388751,0.002619245,0.001050281,0.001425436,0.7705444,0.05697758,0.0231444],"genre_scores_gemma":[0.07934804,0.0004311224,0.04400564,0.0005811906,0.0001364227,0.0006449248,0.8695239,0.0006586557,0.004670117],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.02398653,"threshold_uncertainty_score":0.04769379,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06303050850121361,"score_gpt":0.3912276692429181,"score_spread":0.3281971607417045,"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."}}