{"id":"W4385573045","doi":"10.18653/v1/2022.emnlp-main.115","title":"Enhancing Self-Consistency and Performance of Pre-Trained Language Models through Natural Language Inference","year":2022,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institute for Advanced Research","keywords":"Computer science; Consistency (knowledge bases); Inference; Natural language; Natural language processing; Artificial intelligence; Language model; Natural (archaeology); Cognitive science; Psychology; History; Archaeology","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":[],"consensus_categories":[],"category_scores_codex":[0.0002343513,0.0001094878,0.0001601327,0.00005438919,0.0001476296,0.00003489458,0.0005132254,0.00002249078,0.00003482745],"category_scores_gemma":[0.00001954384,0.00009948073,0.00003355627,0.0002029787,0.00002234309,0.0006625786,0.0006316889,0.0001858657,0.000001061895],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003374348,"about_ca_system_score_gemma":0.0000786128,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002440729,"about_ca_topic_score_gemma":0.00003705335,"domain_scores_codex":[0.9989161,0.00005023225,0.0002390328,0.0003051004,0.0002775452,0.0002119558],"domain_scores_gemma":[0.9993382,0.00007895173,0.00008211889,0.0004343766,0.00003343345,0.00003288449],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00007278071,0.0003456028,0.003405244,0.0007339868,0.0001379395,0.00008680129,0.5084163,0.02480408,0.08178181,0.2703992,0.00008058237,0.1097357],"study_design_scores_gemma":[0.00024449,0.00009291255,0.0005445238,0.00001603301,0.000005398508,0.00002771809,0.001619043,0.9868246,0.009811275,0.0006396899,0.00001337259,0.0001608997],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.843204,0.0007821462,0.1527324,0.00009280888,0.0001247431,0.0001183987,0.000001850949,0.0001717952,0.00277188],"genre_scores_gemma":[0.8948165,0.00001298382,0.1044906,0.0002084268,0.0000171228,0.00001504836,0.000001486219,0.000005376943,0.0004325201],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9620206,"threshold_uncertainty_score":0.4056707,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01190846749545949,"score_gpt":0.2489932632583665,"score_spread":0.2370847957629071,"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."}}