{"id":"W3211439810","doi":"10.18653/v1/2021.emnlp-main.835","title":"Types of Out-of-Distribution Texts and How to Detect Them","year":2021,"lang":"en","type":"article","venue":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"York University","keywords":"Computer science; Categorization; Artificial intelligence; Calibration; Natural language processing; Data mining; Statistics; Mathematics","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":[0.0007588257,0.0001359033,0.0003335481,0.00006521167,0.00004254384,0.00008310398,0.0006012589,0.0000874218,0.000005174188],"category_scores_gemma":[0.002204647,0.00009509638,0.00006016432,0.0007099042,0.00007426301,0.0002290641,0.0004827747,0.0002956404,2.720384e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003300538,"about_ca_system_score_gemma":0.0001132238,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004461249,"about_ca_topic_score_gemma":0.000002834814,"domain_scores_codex":[0.9987881,0.00007261799,0.0002790832,0.0003694251,0.0002969694,0.0001938513],"domain_scores_gemma":[0.9988232,0.000199549,0.0002552066,0.0001717785,0.0005013617,0.00004886835],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00001725014,0.00002447564,0.000826182,0.0002549466,0.000008080502,0.000001286183,0.00330253,0.000005873067,0.1962206,0.003531967,0.00001048644,0.7957963],"study_design_scores_gemma":[0.0002341944,0.00005779259,0.001770415,0.0009370066,0.00001743144,0.00001687811,0.001105645,0.05909836,0.9247481,0.01177516,0.00004183721,0.0001971791],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5934435,0.004173688,0.3932896,0.006707844,0.0003993262,0.0003216608,0.00000781529,0.00004340883,0.001613192],"genre_scores_gemma":[0.6291451,0.00000731649,0.3707127,0.0000510056,0.00001708511,0.000003390461,2.642273e-7,0.000003796457,0.00005938558],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7955992,"threshold_uncertainty_score":0.3877918,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0736749441821645,"score_gpt":0.3899331424867225,"score_spread":0.316258198304558,"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."}}