{"id":"W4287120901","doi":"10.1162/tacl_a_00492","title":"Generate, Annotate, and Learn: NLP with Synthetic Text","year":2022,"lang":"en","type":"article","venue":"Transactions of the Association for Computational Linguistics","topic":"Topic Modeling","field":"Computer Science","cited_by":37,"is_retracted":false,"has_abstract":true,"ca_institutions":"Google (Canada)","funders":"Defense Advanced Research Projects Agency","keywords":"Computer science; Artificial intelligence; Transformer; Natural language processing; Classifier (UML); Machine learning; Labeled data; Task (project management); Distillation; Language model","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":[],"consensus_categories":[],"category_scores_codex":[0.0002956139,0.00007284671,0.0001048999,0.0000623304,0.0005657605,0.00004158121,0.0002870388,0.0000224848,0.000005390613],"category_scores_gemma":[0.0001709704,0.00006428565,0.00005468996,0.0002463946,0.00002223235,0.00003780467,0.00003140102,0.0001269165,6.010141e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001336795,"about_ca_system_score_gemma":0.000121653,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002072119,"about_ca_topic_score_gemma":0.000006446199,"domain_scores_codex":[0.9990729,0.00006746583,0.0002041494,0.0001747528,0.0003644997,0.0001162437],"domain_scores_gemma":[0.998769,0.0004054737,0.0002118887,0.0001613785,0.0004274392,0.00002485121],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000007867027,0.00004727713,0.0005514655,0.00001119137,0.00004281405,1.68352e-7,0.0002747251,0.9435134,0.000009539143,0.05448517,0.00007760255,0.0009788168],"study_design_scores_gemma":[0.0004763199,0.0001083037,0.001067968,0.000007115237,0.00004526674,0.000007294013,0.00005647352,0.9713184,0.0001039498,0.01660016,0.01008966,0.0001190367],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003172662,0.00004384595,0.9945679,0.000822044,0.0005367109,0.0002120451,0.00009536176,0.00004467748,0.0005047997],"genre_scores_gemma":[0.9206119,0.000001656359,0.07821162,0.00009229883,0.0000415479,0.00003213283,0.000007652599,0.000008229345,0.0009929793],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9174392,"threshold_uncertainty_score":0.4351429,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01282966128692613,"score_gpt":0.2278401156433895,"score_spread":0.2150104543564634,"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."}}