{"id":"W4312261477","doi":"10.1109/cvpr52688.2022.00517","title":"Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality","year":2022,"lang":"en","type":"article","venue":"2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":185,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Principle of compositionality; Computer science; Set (abstract data type); Task (project management); Artificial intelligence; Natural language processing; Language model; Field (mathematics); State (computer science); Programming 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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0006912344,0.0003377423,0.0003678318,0.0002647771,0.001028188,0.0005861594,0.0005260672,0.00007798632,0.0001531096],"category_scores_gemma":[0.00001829564,0.0003329201,0.0001021569,0.0002545267,0.00007934779,0.0003878876,0.0005616833,0.0005084338,0.00003028885],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007116452,"about_ca_system_score_gemma":0.000052477,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001580945,"about_ca_topic_score_gemma":0.00001398406,"domain_scores_codex":[0.9972237,0.0004038999,0.0004622325,0.001033601,0.0005248436,0.0003517469],"domain_scores_gemma":[0.9983299,0.0005113318,0.0002563226,0.0004847668,0.0002175214,0.0002001943],"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":[0.0001212503,0.0005931449,0.0003029587,0.0001873852,0.00004934161,0.00002555839,0.004028605,0.002944027,0.003209634,0.007885403,0.001020152,0.9796326],"study_design_scores_gemma":[0.001195047,0.0009089281,0.002789415,0.0001406432,0.00001837397,0.00006951799,0.00007815531,0.9791534,0.0001214669,0.01443998,0.0006448157,0.0004402839],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2280766,0.00004746883,0.7679021,0.00194782,0.0004727478,0.0007839287,0.0001853678,0.0002374167,0.0003465938],"genre_scores_gemma":[0.973434,0.00002601114,0.02317881,0.002361622,0.0002028375,0.000329462,0.0003895497,0.00002894611,0.00004871916],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9791923,"threshold_uncertainty_score":0.9999123,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04285904995147653,"score_gpt":0.3168681275049487,"score_spread":0.2740090775534722,"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."}}